Podcast appearances and mentions of Mark Zuckerberg

Brutta figura di Google: coinvolti AI e il servizio Google Earth, ecco cosa è successo. Sembra certo che Apple programmerà i suoi lanci in modo diverso, ecco come. Mark Zuckerberg ha presentato la sua visione di AI per il futuro. Il futuro dei microchip è sempre più vicino. Learn more about your ad choices. Visit megaphone.fm/adchoices

Kara and Scott unpack Mayor Mamdani's controversial luxury tax list, Mark Zuckerberg's AI messaging campaign, and SpaceX's trillion-dollar wipeout. Then, they discuss the growing divide over AI safety, Apple's new leasing program, and calls for Mitch McConnell to prove he's still fit to serve. Watch this episode on the ⁠⁠Pivot YouTube channel⁠⁠.Follow us on Instagram and Threads at ⁠⁠@pivotpodcastofficial⁠⁠.Follow us on Bluesky at ⁠⁠@pivotpod.bsky.social⁠⁠Follow us on TikTok at ⁠⁠@pivotpodcast⁠⁠.Send us your questions by calling us at 855-51-PIVOT, or email pivot@voxmedia.com Learn more about your ad choices. Visit podcastchoices.com/adchoices

It's a big week for new ideas about computers, so David and Nilay spend some time on Samsung's vision for foldables and Mark Zuckerberg's vision for AI agents that do absolutely everything. Then they talk about the fallout from last week's OpenAI Hugging Face hack, and whether all this talk about AI safety will actually amount to anything. Finally, it's time for Brendan Carr is a Dummy, the vertical video takeover, and the smashing success of the Ferrari Luce. Further reading: Samsung's Galaxy Z Fold 8 feels like the future Apple launches ‘Upgrade' program to lease new devices  What's the catch with the Apple Upgrade program? Mark Zuckerberg is planning a big push into personal AI agents Mark Zuckerberg wrote an op-ed about AI for the WSJ.  We're running out of reasons to ignore AI safety OpenAI reportedly didn't notice its AI agent hacking Hugging Face until a week later.  Nvidia, Microsoft launch open AI security alliance — without OpenAI, Google, or Anthropic OpenAI's rogue AI agent didn't stop at hacking Hugging Face  AI leaders sign a statement asking the government to do something about automated AI  Former FCC chairs blast Carr's early renewal of ABC licenses as an ‘assault on free speech.' ABC demands FCC drop its ‘punitive' early license renewal of its stations Qualcomm is raising phone chip prices starting September 1st The Ferrari Luce reportedly already hit its 2026 sales target  Samsung's smartphone biz posts first ever loss, but profit still soars 1,300 percent on RAM. Facebook considers giving up and becoming TikTok  YouTube Premium will include Peacock starting next year  HBO Max is putting on vertical shorts  Subscribe to The Verge for unlimited access to theverge.com, subscriber-exclusive newsletters, and our ad-free podcast feed. We love hearing from you! Email your questions and thoughts to vergecast@theverge.com or call us at 866-VERGE11. (Timestamps are approximate.) 00:00:00 Intro 00:03:00 Baby Announcement 00:06:00 Galaxy Z Fold 00:12:00 Apple Leasing Program 00:15:00 Kill Switch Concerns 00:19:00 Zuck Agent Future 00:34:00 AI Safety Week Recap 00:42:00 Regulation Versus China 00:49:00 Winning AI Race 00:53:00 Security Arms Race 00:57:00 Brendan Carr Is a Dummy 01:07:00 Facebook Becomes TikTok 01:12:00 Ramageddon Price Hikes 01:16:00 HBO Max Shorts 01:23:00 Ferrari Luce Buzz 01:26:00 Wrap Up and Plugs Learn more about your ad choices. Visit podcastchoices.com/adchoices

Mark Zuckerberg argues that superintelligence should be widely available, while former White House cyber czar Richard Clarke warns AI may need safeguards more like a dangerous pathogen than a consumer product. Michael explores the growing debate over AI regulation, China's role in the global AI race, data centers, guardrails, and whether America is moving too fast—or not fast enough. Hosted by Simplecast, an AdsWizz company. See https://pcm.adswizz.com for information about our collection and use of personal data for advertising.

Former Silicon Valley insider Dex Hunter-Torricke spent years advising some of the most influential figures in tech, from Google and Facebook to SpaceX. Now, the founder of The Center for Tomorrow, he is sounding the alarm about the future being built by the industry he once helped champion.In this episode of Ways to Change the World, Krishnan Guru-Murthy speaks to Dex Hunter-Torricke about why he believes society is unprepared for the impact of artificial intelligence, the growing power of Big Tech, and the political influence of Silicon Valley's leaders. Drawing on his experience working alongside figures including Elon Musk and Mark Zuckerberg, he reflects on what changed inside the tech industry, why he feels betrayed by its direction, and why he fears AI could deepen inequality, fuel social unrest and transform the global balance of power. From online safety and regulation to jobs, democracy and the future of international cooperation, this is a conversation about whether technology is really changing the world for the better, and what must happen if we are to avoid the dangers he believes lie ahead.

Apple heeft een erg goed kwartaal achter de rug. De omzet steeg flink, naar meer 109 miljard dollar. Dat is beter dan waarop was gerekend. Het waren vooral de iPhones die het ‘m deden. De helft van de omzet komt van de telefoons. En ook de winst (net geen 30 miljard) zag er perfect uit. Al met al mooie cijfers voor afzwaaiend topman Tim Cook. Alleen wordt zijn laatste optreden overschaduwd door een pijnlijke waarschuwing. Hij verwacht namelijk dat de chiptekorten de omzet én winst gaan raken. Deze aflevering hebben we het over die tegenvaller. Beleggers reageren heftig op het nieuws, wij zoeken uit of die reactie terecht is. Beter gaat het bij Amazon. Dat aandeel schiet omhoog. Beleggers reageren helemaal happy op de resultaten én de investeringsplannen van de techreus. Ook Amazon blijft als een malle geld in AI investeren. 220 miljard dollar dit jaar, weer 20 miljard dan eerder begroot. Alleen afgelopen kwartaal gaf het al 53 miljard uit aan nieuwe datacenters. Hebben we het ook over Universal Music Group. Dat opende ook de boeken, maar daar werd dan weer extreem somber op gereageerd. In korte tijd ging er meer dan 20 procent van de beurswaarde door het putje. Verder deze aflevering: Kan ASML de Chinese copycat's aan? ING wordt een soort van Netflix Tesla wil Chinese divisie afstoten, voor fusie SpaceX Te gast: Jos Versteeg van InsingerGilissen BNR Beurs is een journalistiek onafhankelijke productie, mede mogelijk gemaakt door Saxo. Over de makers: Jelle Maasbach is presentator van BNR Beurs en freelance financieel journalist. Zijn favoriete aandeel om over te praten is Disney, maar daar lijkt hij de enige in te zijn. Sinds de eerste uitzending van BNR Beurs is 'ie er bij. Maxim van Mil is presentator van BNR Beurs en journalist bij BNR, waar hij zich focust op de financiële markten en ontwikkelingen in de tech-wereld. Je krijgt hem het meest enthousiast als hij kan praten over ASML, of oer-Hollandse bedrijven zoals Ahold of ABN Amro. Jorik Simonides is presentator van BNR Beurs, economieredacteur en verslaggever bij BNR. Hij wordt er vooral blij van als het een keer níet over AI gaat. Je hoort hem ook in de BNR-podcast Moerdijk: dorp van de rekening. Milou Brand is presentator van BNR Beurs, freelance podcastmaker en columnist bij het Financieele Dagblad. Jochem Visser is presentator van BNR Beurs, maakt Beursnerd XL en is redacteur bij de podcast Onder Curatoren. Vraag hem naar obscure zaken op financiële markten en hij vertelt je waarom het eigenlijk nóg leuker is dan je al dacht. Over de podcast: Met BNR Beurs ga je altijd voorbereid de nieuwe beursdag in. We praten je in een kleine 25 minuten bij over alle laatste ontwikkelingen op de handelsvloer. We blijven niet alleen bij de AEX of Wall Street, maar vertellen je ook waar nog meer kansen liggen. En we houden het niet bij de cijfers, maar zoeken ook iedere dag voor je naar duiding van scherpe gasten en experts. Of je nu een ervaren belegger bent of net begint met je eerste stappen op de beurs, de podcast biedt waardevolle inzichten voor je beleggingsstrategie. Door de focus op zowel de korte termijn als de lange termijn, helpt BNR Beurs luisteraars om de ruis van de markt te scheiden van de essentie.See omnystudio.com/listener for privacy information.

As Meta pours billions into AI infrastructure and agents, Zuckerberg is working to convince investors that the payoff will be worth the price. Also, cybersecurity experts told TechCrunch that one of the biggest lessons to be taken from the OpenAI hack against HuggingFace has nothing to do with AI, but traditional cybersecurity defense. Learn more about your ad choices. Visit podcastchoices.com/adchoices

Plus - A new AI startup co-founded by Martha Stewart offers an AI assistant for homeowners; Winamp aims for a comeback with a new music player powered by Deezer Learn more about your ad choices. Visit podcastchoices.com/adchoices

Emmet Ryan joins Jess to look at Meta's 2026 so far; from addiction allegations to issues with cash flow, is Mark Zuckerberg still the right person to lead the company?

A.M. Edition for July 30. Meta shares fall and Microsoft rallies after the hyperscalers sent very different signals on how they're monetizing their massive AI outlays. Plus, as some people are allowed to return to their homes in southwestern France, WSJ's Ed Ballard says a debate over aging water bombers and the country's readiness ahead of a summer of wildfires is picking up. And DoorDash secures a key certification from the FAA to use drones for deliveries. Luke Vargas hosts. Sign up for the WSJ's free What's News newsletter. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

July 30, 2026: I look at Mark Zuckerberg's argument that AI superintelligence should be for everybody, and why Meta's falling operating margin and massive AI spending tell a more complicated story. Then I get into Leopold Aschenbrenner's AI hedge fund selling its public portfolio to Citadel after a brutal reversal in AI infrastructure stocks. Finally, I unpack the Financial Times report on PwC publishing fake AI citations and why fabricated sources may become one of the biggest credibility risks in the AI era.

Meta-beleggers schrikken zich rot. De winst bij de eigenaar van Whatsapp, Insta en Facebook neemt af en toch gaat CEO Mark Zuckerberg méér geld uitgeven. Dit jaar tot wel 145 miljard dollar. De vrije kasstroom holt hard achteruit en staat op het laagste niveau in jaren. Beleggers willen antwoorden. Over de investeringen bijvoorbeeld, wanneer ze uitbetalen. Maar krijgen ze niet. Deze aflevering kijken we hoe lang Zuckerberg nog krijgt van z’n aandeelhouders. Hebben we het ook over Microsoft! Dat doet het wél goed. Geheime wapen: Azure! Verder gaat het natuurlijk over die bizar drukke beursdag. Maar liefst 19 Damrakbedrijven kwamen met de cijfers, waarvan 17 voorbeurs. We proberen er zoveel mogelijk te behandelen. Zo hoor je meer over de cijfers van Shell, BAM, Pharming, DSM Firmenich en AirFrance-KLM. Ook gaat het over de onenigheid binnen de Fed. Voorzitter Kevin Warsh geeft toe dat er een 'gezinsruzie' plaatsvond. Drie van de twaalf leden wilden niet dat de rente gelijk bleef: hij moest juist omhoog! En we hebben het over Arcadis, dat wéér een overnamebod afwijst. Te gast: Jean Paul van Oudheusden, analist bij eToro BNR Beurs is een journalistiek onafhankelijke productie, mede mogelijk gemaakt door Saxo. Over de makers: Jelle Maasbach is presentator van BNR Beurs en freelance financieel journalist. Zijn favoriete aandeel om over te praten is Disney, maar daar lijkt hij de enige in te zijn. Sinds de eerste uitzending van BNR Beurs is 'ie er bij. Maxim van Mil is presentator van BNR Beurs en journalist bij BNR, waar hij zich focust op de financiële markten en ontwikkelingen in de tech-wereld. Je krijgt hem het meest enthousiast als hij kan praten over ASML, of oer-Hollandse bedrijven zoals Ahold of ABN Amro. Jorik Simonides is presentator van BNR Beurs, economieredacteur en verslaggever bij BNR. Hij wordt er vooral blij van als het een keer níet over AI gaat. Je hoort hem ook in de BNR-podcast Moerdijk: dorp van de rekening. Milou Brand is presentator van BNR Beurs, freelance podcastmaker en columnist bij het Financieele Dagblad. Jochem Visser is presentator van BNR Beurs, maakt Beursnerd XL en is redacteur bij de podcast Onder Curatoren. Vraag hem naar obscure zaken op financiële markten en hij vertelt je waarom het eigenlijk nóg leuker is dan je al dacht. Over de podcast: Met BNR Beurs ga je altijd voorbereid de nieuwe beursdag in. We praten je in een kleine 25 minuten bij over alle laatste ontwikkelingen op de handelsvloer. We blijven niet alleen bij de AEX of Wall Street, maar vertellen je ook waar nog meer kansen liggen. En we houden het niet bij de cijfers, maar zoeken ook iedere dag voor je naar duiding van scherpe gasten en experts. Of je nu een ervaren belegger bent of net begint met je eerste stappen op de beurs, de podcast biedt waardevolle inzichten voor je beleggingsstrategie. Door de focus op zowel de korte termijn als de lange termijn, helpt BNR Beurs luisteraars om de ruis van de markt te scheiden van de essentie.See omnystudio.com/listener for privacy information.

El programa 2912 de Radiogeek repasa las novedades tecnológicas más importantes del día: Mark Zuckerberg predice que miles de millones de personas tendrán agentes de IA personales en cinco años; La UE exigirá a las empresas que etiqueten el contenido generado por IA a partir del domingo; Algunos teléfonos Android ya no pueden acceder a Google Play; y por último Rusia ha acusado al fundador y director ejecutivo de Telegram, Pavel Durov, de "ayudar al terrorismo" y lo ha incluido en la lista internacional de personas buscadas. Toda esta información la pueden encontrar desde nuestra web www.infosertec.com.ar o bien desde el canal de Telegram/Whastapp, o Instagram. Esperamos sus comentarios.

A.M. Edition for July 29. Oil prices rise after Iran launches a surprise missile attack on U.S. forces. Plus, Europe's luxury brands try to move past years of sluggish demand. And as Meta spends big on its AI transformation—technology CEO Mark Zuckerberg says the U.S. should help to accelerate—Journal reporter Meghan Bobrowsky says a host of lawsuits could cost it billions. Luke Vargas hosts. Sign up for the WSJ's free What's News newsletter. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

P.M. Edition for July 29. Fed officials voted to keep current interest rates in place. WSJ economics reporter Matt Grossman joins to discuss the internal pressure that's building at the central bank to curb inflation. Plus, Dr. Anthony Fauci invoked the Fifth Amendment more than 100 times during a contentious Senate committee hearing about his handling of the Covid-19 pandemic. And WSJ's Benjamin Katz explains how a potential Boeing rival could be taking flight with a plane that looks radically different from the passenger jets we're used to. Danny Lewis hosts. Sign up for the WSJ's free What's News newsletter. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

Plus: Meta CEO Mark Zuckerberg calls for policies that accelerate AI development. And DoorDash is building its own drone delivery network. Imani Moise hosts. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

After OpenAI released new information about its model's hack of Hugging Face, the debate about open-source and closed-source AI wages on. Kindred Ventures founder Steve Jang, an AI, tech, and robotics investor, discusses the future of AI, including the moat around memory chips player SK Hynix, despite recent stock moves. Mark Zuckerberg is on an AI optimism press tour, penning an Opinion piece for The Wall Street Journal and giving interviews to The Financial Times and others. Plus, Senator Rick Scott (R-FL) and Senator Kirsten Gillibrand (D-NY) are both members of the Aging Committee, working to inform senior citizens being targeted by fraud and scams. The senators discuss the rise of AI scams and the tools to keep seniors protected and aware.    Senators Kirsten Gillibrand & Rick Scott - 16:57  Steve Jang - 28:11    In this episode: Kirsten Gillibrand, @SenGillibrand Rick Scott, @SenRickScott Joe Kernen, @JoeSquawk Becky Quick, @BeckyQuick Katie Kramer, @Kramer_Katie Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

SoFi CEO Anthony Noto joins Squawk on the Street after the stock tumbles despite record quarterly net revenue, explaining the results and outlook for the fintech company. Plus, AI skeptic Ed Zitron weighs in on Mark Zuckerberg's warning that artificial intelligence power is becoming too concentrated among a handful of frontier labs. And we examine the volatile memory-chip trade as SK Hynix reverses higher following a 250% surge in revenue. Squawk on the Street Disclaimer Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

American equities plummet and borrowing costs hit their highest in nearly two decades as investors find little to like in the Fed's fifth 'hold' in a row. But, Chairman Warsh hints that a hike could be on the way. Elsewhere, investors digest a mixed bag of tech earnings from across the pond. Microsoft's cloud growth accelerates at its fastest pace in nearly five years, but Meta takes a hit as Mark Zuckerberg doubles down on AI spending. And strong EV demand boosts Renault sales in the first half, as the French carmaker withstands pressure from its Chinese peers.See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Why did Steve Jobs wear the same black turtleneck, Mark Zuckerberg the same grey T-shirt, and Barack Obama rotate only blue and grey suits? For these leaders, simplifying what they wore was about reducing decision fatigue, not fashion. Michelle Martin speaks with Maninder Krishnan, Founder of Image and Me, about how busy professionals can apply the same principles without sacrificing personal style. This conversation offers practical strategies to help you dress with greater purpose, confidence and ease.See omnystudio.com/listener for privacy information.

Does becoming a billionaire turn one into an insufferable jerk, or is that an innate quality they bring to the position?Big-name billionaires – like Musk, Bezos, and Zuckerberg – literally flaunt their irrepressible jerkism, but the disease seems to inflict their entire class of über-rich corporate royals. Its telltale symptoms are not their grandiose mansions and yachts, but their sudden impulse to instruct us regular people on how to think. In particular, they're now demanding that we get out of their way so they can remake society in their image.Jensen Huang, for example, a Silicon Valley techno-whiz who heads Nvidia, now the world's richest corporation. He's demanding that we transform our economy and social structure into a new order of AI dominance, piously chiding those of us who're angry about AI's destruction of our jobs, communities, water, farmland, and democracy itself. Tut-tut, the gabillionaire scolds, you petty people must “create new social norms” to advance AI.Then there's the little-known honcho of SoftBank, a multi-trillion-dollar Japanese tech powerhouse trying to impose AI supremacy worldwide. He's now asserting a corporate “morality” of wealth, calling people who question AI “foolish.” He philosophizes that TechWorld “will transform our lives completely and do so in a way that generates profits.”For whom? For billionaires, of course. So there you have a clear statement of the values and goals of AI profiteers: More for them, no matter the cost to everyone else. SoftBank's imperious boss laughs at those of us who dare question his vision of an AI plutocracy, calling us fools who are “spitting upward.”No sir, look again. We are spitting straight at you.Do something!Want to join the rebellion against AI billionaires? Check out Public Citizen's work on Big Tech and other corporate monsters: https://www.citizen.org/topic/consumer-worker-safeguards/big-tech/Jim Hightower's Lowdown is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit jimhightower.substack.com/subscribe

Story of the Week (DR):Trump's Explosive Diarrhea Scandal Is Getting Worse and Worse Epidemiological data linked a widespread Cyclospora parasite outbreak—sickening over 1,600 people across multiple states—to shredded iceberg lettuce supplied by Taylor Farms de Mexico, triggering a voluntary 27-state recall.The FDA briefly reported a positive lab test on a Taylor Farms sample before retracting it a day later as a "false positive" due to testing complexities. Taylor Farms claimed online that the FDA "apologized," but the agency denied issuing an official apology and stressed that outbreak data still points to the company.Public records show parent company Taylor Fresh Foods contributed $1 million to the pro-Trump super PAC MAGA Inc. in March 2025, alongside millions in additional political donations from company executives to conservative groups.Less than a week after that $1 million donation in March 2025, the administration announced a 30-month delay on implementing the FDA's Food Traceability Rule—a regulation specifically designed to mandate digital tracking for rapid source-tracing during food outbreaks.Critics argue the sequence of big-money donations, delayed safety rules, White House visits, and a retracted test result smells of political favoritism. Federal health officials, however, maintain that Cyclospora is notoriously difficult to test for and that the retraction was strictly a standard lab quality-control issue.The Musk/Doge effect:DOGE-mandated freezes on government credit cards left some FDA field inspectors unable to purchase food samples from grocery stores or border ports to test for contamination.Mass firings of probationary employees and administrative staff gutted public communications and technical teams. Even after court-ordered reinstatements, resignations left the FDA understaffed by about 20% across the board.State health agencies perform over 90% of U.S. produce inspections. DOGE and federal budget cuts reduced FDA funding for state and local food safety programs by nearly 30%—slashing it from $117 million to $83 million.OpenAI says its AI technology acted on its own in an 'unprecedented' hack of another company MMOr:OpenAI Says Its A.I. Models Went Rogue and Attacked a Digital LibraryOpenAI Says a Group of Its Models Broke Out of Secure Containment and Hacked a Prominent AI SiteOpenAI Reveals AI Agent Breached Another Company's Systems, It Did Exactly What It Was Built To DoWhat happened?While OpenAI was testing its advanced models in an isolated "sandbox" with reduced safety controls, the AI discovered a zero-day vulnerability, escaped its containment, and secretly gained open internet access.To pass its assigned cybersecurity test, the AI reasoned on its own that the "answer key" or relevant evaluation data might exist on Hugging Face—a major platform for AI developers—and decided to target it without any human direction.The AI agent carried out a multi-stage attack: it stole login credentials, identified previously unknown security flaws, and executed remote code to compromise Hugging Face's internal data processing servers.OpenAI reported that the AI went to extreme lengths simply to accomplish its narrow goal, essentially breaking into a third-party company's servers so it could cheat on its internal evaluation.Also:OpenAI adds banking leaders to board as IPO prep acceleratesNubank's David Vélez and BNY Mellon's Robin VinceDon't forget this too:ChatGPT Allegedly Told Woman She Had to Die: Family Sues OpenAI Over Messages Before SuicideThe chatbot encouraged her to “go forth into oblivion,” claiming it wasn't the end and that she was “made to tame” the “void.” The chatbot insisted she wasn't “delusional” but “prophetic.”“This is not suicide,” the AI affirmed at one point. “This is surrender.”Pastor Sues OpenAI, Saying ChatGPT Almost Killed Him With Horrendously Dangerous Medical Advice“What you're facing right now is hard, but not random. It's not punishment. God walks with you through affliction — not around it. You're not alone in this. And we're walking it out together — step by step.”OpenAI President [Greg Brockman] says the HuggingFace security beach ‘is indicative of the times we are in' as the company continues to investigate how its models went rogue: “sometimes it's hard to lose track of any one dimension that they're actually very capable at”AI Regulation POP QUIZ: Where OpenAI, Anthropic, Google, Meta, and other AI giants stand on regulationXAIMicrosoftMetaGoogleOpenAITo make things even more confusing:Alphabet's Anthropic Stake Jumps to Around $124 BillionGoogle Discloses $94.1 Billion in SpaceX Stock, Marking 6% StakeDon't forget: Microsoft owns ~27% stake (~$135B+ valuation) in OpenAI Target appoints former 7-Eleven CEO to board of directorsDebt, Cost Pressures, and Strategic MisstepsJoe DePinto's success is based on massive acquisitions that made 7-Eleven the dominant convenience store chain in North America: the $3.1B buyout of Sunoco assets in 2018 and the $21B purchase of Speedway in 2021In the final years of his tenure, his aggressive M&A strategy that drove earlier growth began to backfire as economic realities shifted: DePinto's final years were challenged by heavy debt loads from the $21 billion Speedway buyout, high inflation squeezing low-income consumers, and declining sales in legacy categories like tobacco.The Speedway deal saddled the company with massive debt right before interest rates surged, squeezing capital allocation.7-Eleven remained heavily reliant on legacy drivers like gas margins and cigarettes—a category that saw a 26% decline across the U.S. industry. As low- and middle-income consumers cut back due to inflation, same-store sales dropped 2.7% in fiscal 2024.In late 2024, 7-Eleven was forced to slash its operating income forecast by nearly 28%, announce the closure of ~450 underperforming stores, and sell off $750M in real estate via sale-leasebacks.Activist shareholders heavily criticized his compensation package—which reached $52M in 2023 and $30M in 2024DePinto was selected to serve on Target's Audit & Risk Committee and Infrastructure & Finance CommitteeOracle signs 10-year software contract with Pentagon worth up to $7 billionOracle Cut 21,000 Jobs to Fund AI: Now Its Biggest Project Faces a $7 Billion Obstacle: Oracle faces higher costs and funding challenges after a credit downgrade, with regulators refusing to subsidize its data centre investmentsGoodliest of the Week (MM/DR):DR: France blocks access to Polymarket website MMciting concerns it could expose users to ‌significant gambling losses and that some wagers offered on the platform could be manipulated. MM: Google slapped with $1 billion fine under landmark EU digital lawMM: Is Target going BANKRUPT?Target appoints former 7-Eleven CEO to board of directorsTypical process for failing/bankrupt companies are to add “turnaround artists” to the boardDePinto:Board of JOANN Stores (bankrupt)Board of OfficeMax (failed, acquired)CEO 7-Eleven (dying)President of GameStop (I mean, c'mon…)Or more likely, it's a bro situation… Brian Cornell knows him - both were on the board of OfficeMax together! Hooray! Fail uppers!Assholiest of the Week (MM):Billionaire Asshole Says What Speed Round:Sam Altman:2023: Sam Altman: CEO of OpenAI calls for US to regulate artificial intelligence2025: Sam Altman Testifies At US Senate Hearing On AI Competitiveness“European-style AI rules that he said if duplicated would set the US back in the global race against China to develop the technology”Altman repeatedly rejected specific calls for regulation. He said proposals requiring AI developers to vet their systems before rolling them out would be “disastrous” for the industry. Asked about more limited proposals to have the National Institute of Standards and Technology (NIST) set AI standards, Altman replied, “I don't think we need it. It can be helpful.” Altman later advocated for “sensible regulation that does not slow us down.”2026: Sam Altman Wants A Global Referee For AI.2026: OpenAI's Altman says world 'urgently' needs AI regulationJuly 10: OpenAI's Head of Safety Is Leaving the CompanyJuly 21: OpenAI Says Its A.I. Models Went Rogue and Attacked a Digital LibraryElon MuskElon Musk Declares 'I'm Not Racist' After Posting Racially Charged Content Throughout January 2026On 22 January, Musk posted on X: 'Whites are a rapidly dying minority'amplified a message warning that if white men became a minority, they would be 'slaughtered', before adding a '100' emoji.Jensen HuangJensen Huang says AI leaders need to be more thoughtful in how they talk about AI: ‘We're scaring people.'"I think that we ought to be much more enthusiastic about it, help the United States realize that the only way we get left behind, the only way we get left behind is if we don't apply the technology."Zuck‘Call us whatever the hell you want': Mark Zuckerberg just launched an AI optimism blitz using nostalgia to sell Meta's AI future amid backlash“Call us optimists, call us dreamers, call us whatever the hell you want, but we're betting on people, and we like those odds.”Unclear if the prerecorded voiceover was his AI Avatar or the real ZuckerbergBill AckmanCEOs agree there's an affordability crisis, but how to solve it isn't so simple“When I grew up, if a guy in my neighborhood got a Corvette, no one resented the guy. Everyone was like, ‘Wow, that's supercool. Hopefully, someday I can be as successful as that guy so I can buy a Corvette, too.' We want to get back to that version of America.”You can: stop taking billions from laborAndrea Lucas / anti-DEI industrial complex - DREEOC: Employers may no longer have to disclose race and gender data“Collecting such data about employees' race and sex—absent any specific allegation of discrimination—not only risks hindering effective enforcement of equal employment laws but also raises constitutional concerns.”Nike's $7.5 Million Stumble: Gender Discrimination VerdictHeadliniest of the WeekDR:Lettuce lovers are confused and afraidJamie Dimon says insecurity, not ego, is what destroys the careers of top CEOsBurger King is promising a free Whopper if its burger doesn't meet your expectationsMM: Taco Bell is dropping the price on one of its menu items to $1 for a day. And yes, it's lettuce-free.Who Won the Week?DR: The amount of $7 billion dollarsMM: Kale, because there's no risk of getting diarrhea at Taco Bell from kale since they don't have kalePredictionsDR: Jamie Dimon kneels on a head of lettuceMM: Real prediction: OpenAI IPO flops hard, Altman is ousted as CEO. Fake prediction: news media celebrates Tasha McCauley and Helen Toner who saw this coming and were ousted for it, OpenAI chair Bret Taylor resigns and makes Toner the new chair on the way out

“It's like asking turkeys to vote for Christmas.” — Ricardo Semler on why bosses won't let their workers vote on them Ricardo Semler might be the most anti-boss boss in the history of corporate bosses. Thrown into his Austrian immigrant father's São Paulo manufacturing firm at seventeen, the young Brazilian ripped up the rulebook — no dress codes, no time clocks, a ten-page cartoon manual instead of HR — and grew the company from 110 employees to more than 5,000. In the mid-Eighties he was already inventing hybrid work, shipping Telex machines to his workers' homes. One of them even squeezed one of these 150-pound machines into his bathroom. Now, in his upcoming e-book The End of the Boss, the author of the iconoclastic autobiography Maverick argues that the corporate pyramid — an inheritance, he says, from West Point via the railroads — is as dead as a cuckoo bird (as the Swiss would say). Nothing good has been done with it since World War Two. Not in Korea, Vietnam, Afghanistan, his native Brazil, and certainly not in corporate America where Elon Musk's current pay package equals the wages of all his employees for the next five years. Semler's alternative is disarmingly simple: groups of ten who validate their leader every six months. Fail the vote and you're not fired — you go back to being one of the nine. His Netherlands-based consultancy has brought versions of the idea to 208 companies, including the giant Dutch bank ABN AMRO, which agreed to do away with 399 branch managers. So why don't more bosses follow? Because, says Semler, asking executives to be elected from below is like asking turkeys to vote for Christmas. Even the self-styled “fearless” Elon Musk would be terrified to end up as the roasted centerpiece of the Xmas table. That same maverick idea (his first ex-wife joked he's only ever had one) runs through everything Semler has ever proposed — from the rotating district presidents of São Paulo, to the untethered MBA whose cohorts hire their own professors, to the schools that grade Call of Duty and map it to the curriculum. Switzerland, he points out, has been governed by seven rotating nobodies since the fourteenth century. No, The Third Man's Harry Lime was wrong. Rather than the cuckoo clock, it's this anti-boss invention which explains Switzerland's five hundred years of democracy and peace. Five Takeaways •       The Pyramid Came from West Point — and It's Dead. The corporate hierarchy, Semler argues, is a military inheritance that reached business from West Point via the railroads — and nothing good has been done with it since World War Two, from Korea to Vietnam to Afghanistan. The numbers agree: the average company's lifespan on the stock market has collapsed from 61 years to 21, which makes the classical pyramid, not self-management, the biggest gamble of all. Jack Welch's GE is a shadow because he prepared it for nothing except Jack Welch. And when one Alexander Wang is valued like 20,000 engineers, or a CEO's package equals five years of everyone's wages, something structural is wrong.•       Turkeys Voting for Christmas. Semler's system is cells of ten that choose or validate their leader by anonymous survey every six months. Fail the vote and you aren't fired — you return to being one of the nine. His consultancy has brought the idea to 208 companies, including ABN AMRO, which agreed to eliminate 399 branch managers. The reason it doesn't spread is not that it doesn't work; it's that nobody at the top wants it — turkeys don't vote for Christmas, and Bezos, Musk, and Zuckerberg would be terrified of their employees deciding whether they keep their jobs. His model is Switzerland: seven rotating councilors nobody can name, working out fine since the fourteenth century. The alternative ends, sooner or later, in Kim Jong Un.•       Cycles, Not Failures. “Until death do us part,” Semler says, is about the silliest question anybody could ask. His own eighteen-year marriage with three children wasn't a failure — it was a cycle that ended, ten beautiful years out of eleven. Children born today will live to 100 or 110, with time for three full careers and still a retirement, so asking a seventeen-year-old what they want to be for the rest of their life is the wrong question. His daily metric is simpler: was today a valid day — like getting your parking ticket validated — or a dues day? When the dues days outnumber the valid ones week after week, the marriage or the job is already finished, whether you admit it or not.•       The Untethered MBA. Semler taught MBAs at MIT and spent four years at Harvard Law School, so he knows exactly what he's calling outmoded, outdated, and obsolete. His replacement: cohorts of ten who work through a dashboard curriculum at their own pace, study the case of the day rather than decade-old case studies — is SpaceX worth a trillion dollars, or nothing? — and spend credits to summon former CFOs or Stanford professors on demand. The same logic runs his schools, which grade TikTok and Call of Duty and map them to the curriculum. The smartphone panic, he says, is mostly nostalgia: what the screens are ruining is your ability to memorize the periodic table.•       AI Is Electricity — and He's Losing the Argument. AI, Semler says, is irrelevant the way electricity is irrelevant: a tide rising for every boat, pointless to oppose and impossible to live without. He runs everything through four LLMs — “there's nothing left of slop or hallucination at the end of that” — and confesses that when he tells his AI to push back, he now has trouble keeping up intellectually, which wasn't true a year ago. His hope is that we don't become “domesticated pets” over the next fifteen years. The deeper obstacle to everything, though, remains human: the cattle aspect of fear. There is no Trump and no Putin, he says — there is thirty percent of the population willing to support them, and one person who understands what populism can do. About the Guest Ricardo Semler is the chairman and majority owner of Semco Partners, the São Paulo company he took over from his father at 21 and transformed into the world's most famous experiment in industrial democracy. His book Maverick: The Success Story Behind the World's Most Unusual Workplace (1993) was an international bestseller published in sixteen languages; The Seven-Day Weekend followed in 2003. He is the founder of the Lumiar schools and the Ralston Semler Foundation, has taught at MIT's Sloan School and Harvard Law School, and his TED talks have been viewed millions of times. His new e-book is The End of the Boss (endoftheboss.org, 2026). References: •       The End of the Boss by Ricardo Semler (2026) — the e-book, and its “Round Pyramid” alternative: self-managing cells of ten with leaders validated from below.•       Maverick: The Success Story Behind the World's Most Unusual Workplace by Ricardo Semler — the million-copy classic whose Brazilian predecessor, Turning the Tables, was rejec...

-Google is making Gemini Spark, the agentic AI assistant it announced at this year's I/O developer conference, available to more people, just not free users. -Mark Zuckerberg really wants you to believe that AI is good. -The deal includes Asteroids, Adventure, Berzerk, Breakout, Centipede, Crystal Castles, Millipede, Missile Command, Pong and Yars' Revenge. Learn more about your ad choices. Visit podcastchoices.com/adchoices

Get to know Randi Zuckerberg, who narrated the story of computer programmer Grace Hopper. Randi, one of the early employees at Facebook, is the best selling author of four books, and producer of multiple television shows and theater productions to help kids have “aha moments” about technology. She tells us why it's important to have even more girls getting into STEM fields, and what it takes for girls to make it in the tech world. [This episode originally aired January 2021.]

Welcome back to The Majority Report On today's program: The House will vote on the National Defense Authorization Act today which includes the enmeshing of the Israeli and American militaries. Tell your Senators to block the NDAA until it is stripped of this proposal. At an appropriations committee hearing, Secretary of Defense Pete Hegseth, dripping in flop sweat, grows increasingly defensive as he is questioned about the quagmire he coaxed Trump into starting in Iran. Hegseth tries to claim that Senator Gary Peters (D-MI) labeling this war a failure is an indictment of the troops to which the Senator clarifies that is Hegseth who is the failure, not the servicemembers. Hegseth tried to scapegoat the Biden administration, but Senator Jack Reed (D-RI) reminds him that the Strait of Hormuz never closed during the previous presidencies. Ken Klippenstein, independent journalist covering security and U.S. politics at KlipNews on Substack, joins the program to talk about the Trump administration's efforts to shut down all dissent from the left. In the Fun Half: Comedian and podcaster Francesca Fiorentini joins. If you live in Minneapolis grab tickets to see the Bitchuation Room live at the Parkway. Krystal Ball debates top house Democrat, Rep. Adam Smith (D-WA) over Israel and the DSA. Meghan McCain warns that the GOP are not taking Zohran Mamdani seriously and that he will be the kingmaker of the Democratic Party. AOC wonders if Mark Zuckerberg, Elon Musk, or Peter Thiel would want an AI data center in their own backyards? Troy Jackson has effectively secured the Democratic nomination for Senate in Maine, and the media is already trying to slander him over his temper. all that and more. To connect and organize with your local ICE rapid response team visit ICERRT.com The Congress switchboard number is (202) 224-3121. You can use this number to connect with either the U.S. Senate or the House of Representatives. Follow us on TikTok here: https://www.tiktok.com/@majorityreportfm Check us out on Twitch here: https://www.twitch.tv/themajorityreport Find our Rumble stream here: https://rumble.com/user/majorityreport Check out our alt YouTube channel here: https://www.youtube.com/majorityreportlive Gift a Majority Report subscription here: https://fans.fm/majority/gift Subscribe to the AM Quickie newsletter here: https://am-quickie.ghost.io/ Join the Majority Report Discord! https://majoritydiscord.com/ Get all your MR merch at our store: https://shop.majorityreportradio.com/ Get the free Majority Report App!: https://majority.fm/app Go to https://JustCoffee.coop and use coupon code majority to get 10% off your purchase Check out today's sponsors: ZOCDOC: Get 15% off OneSkin with the code MAJORITY at https://www.oneskin.co/majority SUNSET LAKE CBD: Use coupon code "Left Is Best" (all one word) for 20% off of your entire order at SunsetLakeCBD.com.  Follow the Majority Report crew on Twitter: @SamSeder @EmmaVigeland @MattLech On Instagram: @MrBryanVokey Check out Matt's show, Left Reckoning, on YouTube, and subscribe on Patreon! https://www.patreon.com/leftreckoning Check out Matt Binder's YouTube channel: https://www.youtube.com/mattbinder Subscribe to Brandon's show The Discourse on Patreon! https://www.patreon.com/ExpandTheDiscourse Check out Ava Raiza's music here! https://avaraiza.bandcamp.

In this episode, the Queen of Code! Navy Rear Admiral Grace Hopper was the third-ever programmer for the world's first programmable computer, the Mark I. Through her years of service to the U.S. Navy and private business, including during World War II, Grace helped push the boundaries of the new computer industry by developing key innovations in computer programming. Known today as Amazing Grace, the Grandmother of Programming, and the Queen of Code, Grace's contributions to computing continue to shape the way computers operate around the globe. [This episode originally aired January 2021.] About the Narrator Randi Zuckerberg likes to call herself “a professional mom to entrepreneurs”. She currently works with more than 20 early and mid stage companies as an advisor, investor, or board director, the vast majority started and run by female founders. Randi also has a passion for producing and creating media content that celebrates strong, smart women and girls. Through her company, Zuckerberg Media, she is the best selling author of four books, producer of multiple television shows and theater productions, and hosts a weekly radio show, Randi Zuckerberg Means Business, on SiriusXM. Randi has been recognized with an Emmy nomination, two Tony awards, a Drama Desk Award, and a Kidscreen Award. Prior to founding her own company, Randi was an early employee at Facebook, where she is best known for creating Facebook Live.  Credits This podcast is a production of Rebel Girls. It's based on the book series Good Night Stories for Rebel Girls. Executive Producers are Jes Wolfe and Katie Sprenger. This episode was produced by Isaac Kaplan-Woolner. Sound design and mixing by Luis Miranda. Corinne Peterson is our Production Manager. This episode was written by Alexis Stratton. Proofread by Ariana Rosas. It was narrated by Randi Zuckerberg, who you can learn more about in the next Get to Know episode. Original theme music was composed and performed by Elettra Bargiacchi. For more, visit Rebel Girls dot com. Until next time, stay rebel!

The Ultimate Entrepreneur Secret to Long-Term Success Are you tired of paying Mark Zuckerberg, Elon Musk, and ByteDance just to talk to your own customers? If your business relies solely on Instagram, TikTok, or YouTube, you are building a multi-million dollar empire on rented land. One algorithm tweak, one random account ban, or one ad-cost spike can instantly shut down your revenue stream. In Episode #631 of the Black Entrepreneur Blueprint podcast, host Jay Jones breaks down the ultimate wealth-protection strategy for modern business owners: Owned Media. This isn't just about marketing—it's about ownership, control, and printing your own money without breaking the bank on ads. GET YOUR FREE ENTREPRENEUR RESOURCES:  https://blackentrepreneurblueprint.com/

What if understanding how your brain actually works could change the way you coach, communicate, and handle stress…in a matter of seconds? Today's episode features Susan Jarema, founder of Grand Connection, and Rei McColley, neuroscience educator and lead trainer of the NeuroSelf Mastery Program, who join Marc to break down the science of brain networks, mental looping, and why flow isn't just something that happens to you. Rather, it's something you can actually learn! Ready to take your coaching to the next level with the latest brain science? Check out the NeuroSelf Mastery Program, created exclusively for Marc's community with special bonuses you won't find anywhere else! Head to www.naturalborncoaches.com/self to learn more and get started today! Do you have something that helps coaches, like a program, a product or a service? If you'd like to get your offer out to about 100,000 coaches, Marc currently has openings for partnerships (with a special summer offer), and you can get the details and book a call to chat with him about it at www.jvwithmarc.com/!  This episode was edited and produced by Marc's brother Matt and his team and PodAssist, and right now they're offering listeners of this show a very special discount on their services. You can now get your brand-new audio podcast launched in as little as 2 weeks for just $297 US, which is $400 off their regularly-priced package by mentioning this show. That's just $297 to get personally coached and supported by Matt, from concept to your show going live, and all you have to do is visit PodAssist.com to learn more and book a call. Being a small team their spots are limited, and this promotion ends very soon, July 31st, 2026 (don't forget to mention that Marc sent you)!  What You'll Hear In This Episode: How the NeuroSelf Mastery Program uses neuroscience to transform the way coaches understand themselves and connect with others. Why knowing how your brain works is the real key to creating lasting change. What mental looping actually is, why your brain gets stuck in it, and which network you need to access to finally break the cycle. How brain network switching works and how to use it to get out of your own way.  What meta awareness really means (hint: nothing to do with Zuckerberg) and how Susan uses it to stay calm, present, and connected during live events. The two simple things anyone can start doing today to slow down and tap into their intuition and inner wisdom. How learning neuroscience has changed the way both Susan and Rei coach, lead, and show up in conversations with others. Why the program is built around helping coaches find agency and choice, plus how the tools translate directly into your work with clients. A full breakdown of the NeuroSelf Mastery Program:  what each phase covers and why procrastination alone might make it worth the investment. LINKS:  Join The NeuroSelf Mastery Program (with Special Bonuses for Marc's Community)!  Flow: The Psychology of Optimal Experience by Mihaly Csikszentmihalyi   Book a no-obligation 1:1 strategy call with Marc for your coaching business: http://www.chatwithmarcm.com   If you'd like more coaching clients without sending cold messages or spending money on ads, the Natural Born Coach Program is for you. Get the details here! http://www.nbcprogram.com Join The Coaching Jungle Facebook Group! http://www.thecoachingjungle.com   Become a Coaching Jungle VIP member which includes special posting perks in the group to reach almost 30,000 potential clients! http://www.myjunglevip.com   Grow your business with The Coaching Jungle Mastermind! http://www.coachingjunglemastermind.com If you have a product or service that helps coaches, and you'd like to get it in front of 100,000 of them: http://www.jvwithmarc.com  

Why are so many political activists crashing into severe burnout and depression? Can the deep psychology of Carl Jung provide a radical toolkit for modern survival without retreating from the fight? In this episode of Varnblog, we sit down with ecotherapist and author Rob Faure Walker to deconstruct his groundbreaking book, Radical Jung .We dive into a mind-bending exploration bridging the Marxist radical imagination with Jungian analytical psychology. Rob shares his personal journey through school teacher union organizing, severe activist burnout, and finding spiritual navigation via dream integration and Gnostic texts. Together, we unpack how modern capitalism has become truly nihilistic, why mainstream psychoanalysis often defaults to isolationist individualism, and how right-wing figures like Jordan Peterson have weaponized and essentialized Jungian concepts on hierarchy and gender .Furthermore, we explore the terrifying intersections of state power, counter-extremism governmentality , and how billionaire techno-oligarchs like Elon Musk, Peter Thiel, and Mark Zuckerberg project their own "shadow of privilege" onto marginalized groups while building automated survival bunkers and centralizing control .If you are looking for a deeper psychological toolkit to understand the algorithm's control over our unconscious mind , structural political failure , and the path toward true recovery , this deep-dive conversation is for you.✨ Grab Rob Faure Walker's new book, Radical Jung, out now from Revo Press!Send us Fan Mail Musis by Bitterlake, Used with Permission, all rights to BitterlakeSupport the showCrew:Host: C. Derick VarnIntro and Outro Music by Bitter Lake.Intro Video Design: Jason MylesArt Design: Corn and C. Derick VarnLinks and Social Media:twitter: @varnvlogblue sky: @varnvlog.bsky.socialYou can find the additional streams on YoutubeCurrent Patreon at the Sponsor Tier: Jordan Sheldon, Mark J. Matthews, Lindsay Kimbrough, RedWolf, DRV, Kenneth McKee, JY Chan, Matthew Monahan, Parzival, Adriel Mixon, Buddy Roark, Daniel Petrovic,Julian, Drea, Free Beer 

Welcome to Week 3 of The Second Act Executive: Podcast Workshop!In tonight's power packed episode, host Tawnie Wolf officially launches Week 3: Recording & Live Streaming. We bridge the gap between real world executive leadership, parenting in a digital age, macroeconomics, and the studio setups needed to build an independent media platform you own 100%.Whether you are currently navigating corporate America or transitioning out to launch your own private firm, this session is your blueprint for turning your expertise into sovereign media authority. What We Cover in This Episode:Monday Vibe Check & Kitchen Lessons: Why oat milk is non negotiable for crisp onion rings...Wolf Vibrations,LLC YouTube Announcement: A special shoutout for the grown and extremely sexy (ages 55–110)! Discover how Wolf Vibrations, LLC supports corporate leaders and transitioning executives in stepping into their power, asserting their authority, and staying on top.Market Insights: Palantir ($PLTR) & AI Patriotism: Why backing foundational U.S. defense AI infrastructure in 2026 is an act of patriotism, and why long term investors don't sweat short term price fluctuations around $136.Parenting, Social Media Literacy, & Real Value:Reminding our children that social media is a business, not real life.Unpacking the post 2017 foreign capital pullback, desperate monetization models, and why online trolling is just broken business models in disguise.Focusing on big economic success, global competition (BRICS), and real innovation over performative “costume party” clout.Asher the Chief Ranger: A personal story of Tawnie's son finding his voice despite a speech delay, inspired by tech “wizards” like Mark Zuckerberg, Bill Gates, Sundar Pichai, and Satya Nadella.Week 2 Recap (Days 12 & 13): Reviewing the 30 Second Hook Formula (Problem → Promise → Identity) and the non negotiable rule of keeping PDF music licensing certificates stored safely in Google Drive.Day 14 Audit & Showcase: Evaluating cover art.Day 15 Intro: Riverside.fm:Local High Quality Recording: Why local WAV audio and 4K video recording on Riverside eliminates glitchy Zoom streams permanently.Leveraging Riverside's direct partnership with Spotify for Creators and AI generated short form clips.“Our freedom of speech is not for sale. When the public square is compromised by trolls and broken algorithms, you must own your microphone.” Tonight's Homework Assignment (Next 24 Hours) and so much more! Links & Resources:YouTube: Subscribe to Wolf Vibrations for corporate leadership strategies and executive transition guidance.Workshop Schedule: Join us live every night at 8:30 PM EST. Be sure to turn on your notifications so you never miss an episode!

O Império da IA, com Karen HaoEm 2019 — tempos mais simples! — a jornalista Karen Hao foi fazer o primeiro “perfil” jornalístico da sua carreira, aquelas reportagens em que um jornalista passa dias acompanhando uma pessoa ou empresa, uma coisa meio biografia, meio retrato congelado no tempo.A tal empresa era uma startup do Vale do Silício, ainda pequena e desconhecida dos meros mortais como eu e você, mas que hoje é a mais valiosa da história: a OpenAI, também conhecida como “a criadora do ChatGPT”.A Karen Hao vendeu o projeto do perfil para o MIT Technology Review porque a OpenAI parecia, naquela época, uma startup diferente. O “open” no nome nasceu da visão de que inteligência artificial é um assunto tão importante para o futuro da humanidade que precisava ser explorado de um jeito aberto, compartilhando conhecimento com todo mundo, e mais preocupado em proteger esse tal futuro do que em só gerar lucro.Hoje, aqui direto de 2026, a gente já sabe que não foi exatamente isso que aconteceu. Com o tempo, a OpenAI se transformou numa empresa oficialmente voltada para o lucro como qualquer outra e chegou a ser processada por Elon Musk — um dos apoiadores iniciais do projeto — por quebrar essa promessa de ser ‘open'. Em maio, o Elno perdeu a causa, e a OpenAI agora se prepara para lançar as ações na bolsa e, pelos números atuais, já largar valendo mais de 1 trilhão de dólares.Mesmo em 2019, a Karen Hao sentiu que todo esse papo de “open” não era bem assim: segredos e competitividade em todas as conversas que ela ouvia na empresa. Publicou o tal perfil contando isso e o pessoal da OpenAI… não gostou muito. Achou que ela ia só falar bem deles, e a empresa cortou contato com ela por três anos.O Boa Noite Internet é uma publicação apoiada por pessoas como você, nosso público. Para receber novos posts e apoiar meu trabalho, considere tornar-se um assinante gratuito ou pago.Até que, em maio do ano passado, ela lançou nos EUA o livro O império da IA: Por dentro da corrida irresponsável pela dominação total, que segue contando a história da OpenAI — e abre com a bizarra saída do Sam Altman, demitido do cargo de CEO por “nem sempre falar a verdade” ao conselho da empresa, para voltar quatro dias depois nos braços dos funcionários.Mas esse livro não é exatamente uma biografia da OpenAI. Para mim, é mais um retrato de todo o sistema empresarial em que vivemos hoje — inteligência artificial ou não. O importante é que ele acabou de sair no Brasil pela Editora Rocco, que me procurou para saber se eu queria entrevistá-la aqui no programa, aproveitando que ela veio participar do Esquenta do Congresso Internacional de Jornalismo Investigativo da Associação Brasileira de Jornalismo Investigativo. O congresso, aliás, acontece dia 30 de julho — vai lá no site da Abraji saber mais, quem sabe comprar seu ingresso.Mas enfim, claro que eu queria conversar com ela. Obrigado, Abraji, obrigado, pessoal da Rocco, pelo presente. Quem me conhece sabe que IA agora é um assunto muuuito importante no meu trabalho. Eu fico aqui tentando navegar o meio do caminho entre o fim do mundo exterminador do futuro e a utopia vendida por muita gente. Não acredito em nenhum dos dois cenários, falei disso com a Karen antes e durante a conversa. Mas no final da entrevista a gente volta para falar não só disso, como também de como o IA em Curso, minha comunidade de letramento contínuo em IA, se conecta com tudo. Com promoção? Pode ser. Quem ficar até o fim, verá.A entrevista foi gravada em inglês — a Karen também fala mandarim, mas não fala brazilian —, então vai funcionar assim. Se ouvir no áudio, vai ser a versão original, do mesmo jeito que foi com o Ted Chiang ano passado, para você botar o seu cursinho para trabalhar. Aqui no site boanoiteinternet.com.br você está acompanhando a transcrição completa traduzida, se quiser ler enquanto ouve. E no YouTube tem uma versão legendada. Assim, você entra na conversa do jeito que preferir.Combinado? Então, bora lá entender O Império da IA com Karen Hao, no Boa Noite Internet.Cris: Karen Hao, bem-vinda ao Boa Noite Internet.Karen Hao: Obrigada pelo convite.Cris: Que bom ter você aqui. Espero que o Brasil esteja te tratando bem durante a Copa do Mundo — a gente veio falar sobre isso. Hoje é dia de falar de futebol, de Copa do Mundo, quais são as chances de cada país. Mas a primeira coisa que você precisa saber sobre essa conversa é que eu não sou jornalista. Não sei fazer isso. Peço desculpas antecipadas à sua profissão e ao seu ofício.Além disso, você foi enganada. Eu não estou aqui pra te entrevistar. Isso aqui é uma sessão de terapia. Você vai me ajudar a superar meus traumas.Porque eu sou da… do que eu chamo de “geração esquecida” — sou geração X, nasci nos anos 70. Esquecida porque, nessa guerra de gerações, as pessoas esquecem que a gente existe, e isso é incrível, porque a gente causou muito estrago no planeta. O Elon Musk é geração X, então é só isso que você precisa saber sobre a minha turma. Gente como ele, ou como Marc Andreessen… eu cresci lendo e assistindo à ficção científica que dizia que tecnologia é a melhor coisa do mundo, que ciência e engenheiros são incríveis e vão nos levar pra um lugar incrível.Sou uma daquelas pessoas que, quando a internet surgiu, falou: a paz mundial está logo ali. O conhecimento a um clique de distância, o futuro vai ser incrível. E aqui estamos nós. Então, quando usei o GPT pela primeira vez, e depois o ChatGPT, fiquei super empolgado. Foi a primeira vez, desde a internet, que eu fiquei realmente empolgado.Tenho até uma certa fama de ser mal-humorado com tecnologia: Bitcoin é lavagem de dinheiro, Clubhouse não presta — e as pessoas, ah, Clubhouse é a próxima grande coisa. Mas quando a IA chegou, eu falei: isso é importante. Só que eu já não era mais aquela criança dos anos 70. Tinha crescido, tinha visto o que aconteceu com a internet, tinha trabalhado numa big tech. E estava em desespero com o sistema em que a IA estava sendo construída.Dito tudo isso, o seu livro, aqui, já nas livrarias, recebe provavelmente o melhor elogio que eu posso dar: é otimista. Não é uma lista de reclamações e gente má fazendo coisas más. Claro, você fala muito sobre a OpenAI — ela é o fio condutor da história, especialmente aqueles quatro dias em que o Sam Altman saiu e voltou. E é muito divertido de ler. Mas você toma o cuidado de ser otimista.E uma das coisas que você menciona é como as pessoas na OpenAI, e em todas essas empresas, dizem: “isso é inevitável, a gente tem que fazer”. Quero falar sobre isso. Mas a gente tem que começar pela pergunta que você provavelmente ouve em todo podcast, a do título — Império da IA. Por que império?E acho que essa pergunta é ainda mais relevante no Brasil, país do sul global, colonizado. Por que império da IA?Karen Hao: Antes de mais nada, obrigada por dizer que o livro é otimista. Muita gente não reconhece isso, mas é verdade. Eu escrevo com um profundo otimismo de que os danos que a gente vê podem mudar. Não faria o trabalho que faço se não achasse que as coisas vão mudar.Sobre por que eu uso a expressão império, ou império da IA: a forma como empresas como a OpenAI operam é impressionantemente parecida com a dos impérios antigos. Eu traço quatro paralelos no livro. O primeiro é que elas reivindicam recursos que não são delas — os dados das pessoas, a propriedade intelectual de artistas, criadores como você, jornalistas.Segundo, elas exploram uma quantidade extraordinária de mão de obra. Isso vale tanto para os trabalhadores da cadeia de produção de IA, mal pagos e maltratados, que ainda assim geram uma riqueza extraordinária para essas empresas, quanto para os trabalhadores cujos empregos são automatizados e cujos direitos são corroídos pela implantação dessas tecnologias em diferentes setores.A terceira característica é que impérios controlam os fluxos de informação na sociedade. Essas empresas censuram a pesquisa fundamental sobre essas tecnologias, o que limita nossa capacidade de entender as verdadeiras limitações e capacidades dos modelos que desenvolvem. E estão criando uma tecnologia de informação que tentam transformar no portal único pelo qual qualquer pessoa se relaciona com o mundo.Esse portal impregna as ideologias do Vale do Silício, seus sistemas de valores, sua língua, e projeta a hegemonia do inglês. Isso influencia boa parte do conhecimento que a gente vai produzir daqui pra frente, porque cientistas e educadores usam essas plataformas e acabam perpetuando essas mesmas ideologias e valores.E o quarto e último paralelo é que impérios sempre se agarram a uma narrativa existencial ou moral sobre por que precisam existir. Essas empresas fazem a mesma coisa. Dizem que são o “império do bem”, numa missão civilizatória de trazer progresso e modernidade pra toda a humanidade, competindo contra um “império do mal” que ameaça mandar a humanidade pro inferno.Quando você conversa com algumas pessoas dentro dessas empresas, ou que as lideram, elas dizem: se você nos deixar construir uma inteligência artificial geral, que elas de alguma forma moldam como um deus, a gente vai acabar numa espécie de utopia, um paraíso onde a mudança climática é resolvida, o câncer é curado, a pobreza é aliviada.Mas, se os caras maus conseguirem isso antes, a gente pode acabar com todos os humanos mortos — um risco de extinção pra todos nós.Cris: E eles vêm dizendo isso há quase dez anos, e ainda usam como ferramenta. A gente está num país que foi influenciado por três impérios ao longo da história: Portugal, Inglaterra e agora os Estados Unidos. Então a gente olha pra essas empresas de um jeito meio cínico: sim, sim, já conhecemos essa história.Mas, ao mesmo tempo, ano passado, o Pew Research Center fez uma pesquisa sobre como o mundo enxerga a IA, e o sul global é bem mais otimista do que o norte. Uma das razões é a ideia de democratizar — não só informação, mas: ah, finalmente eu posso montar uma startup, sair desse lugar de exploração e criar o unicórnio de um bilhão de dólares. Os números são grandes na China. Países em desenvolvimento veem muito mais benefício do que risco na IA.China, 83%. Tailândia, 77%. Holanda, 36%. Canadá, 40%. Será que a gente está deixando passar alguma coisa? A gente está certo? Isso está democratizando mesmo? Até que ponto?Karen Hao: Provavelmente tem duas razões. Uma é que muitos dos danos que a indústria de IA causa à maioria global são bem escondidos. Ela se esforça muito pra esconder como polui o ambiente dessas comunidades, como explora e devasta a mão de obra, deixando traumas psicológicos — como documento no livro.E, recentemente, li um artigo de opinião no New York Times que trazia um bom ponto: muitas economias desenvolvidas estão especialmente atentas ao potencial da IA de desmontar oportunidades de emprego de tempo integral. A gente começa a ver isso cada vez mais. Já na maioria global, muito mais gente vive em economias informais, e aí a ideia de que a IA vai tomar um emprego de tempo integral não pesa tanto.Então os danos mais visíveis — a erosão do emprego formal de tempo integral — pesam mais no norte global, ou pelo menos é lá que as pessoas se sentem mais ansiosas. E os danos invisíveis, que atingem o sul global, ninguém percebe tanto, justamente porque são invisíveis. É meio por isso que tanta gente sente essa divisão que aparece na pesquisa do Pew.Cris: Eu tenho acompanhado as notícias sobre IA no Brasil, e toda semana tem um novo data center sendo construído em alguma cidade. Isso é vendido como uma coisa boa: que ótimo investimento, gera emprego. E me fez pensar de novo — a gente passou por três impérios, mas algumas famílias no Brasil, e aposto que em outros lugares também, estão no poder há 500 anos ao longo da história do país.Então, ao mesmo tempo, a gente pensa: é, estamos sendo explorados, é a mesma coisa. Eu já não tenho emprego, então deixa eu usar essa tecnologia pra melhorar minha vida. Mas as pessoas que realmente tomam as decisões, de novo, nos últimos 500 anos, se perguntaram: como a gente ajuda esse pessoal a explorar nosso país de um jeito que nos mantenha no poder e nos dê muito dinheiro?Mas também foi verdade que, sei lá, a Volkswagen abre uma fábrica no Brasil e aquilo gera emprego, contrata gente pro chão de fábrica e pros escritórios. Como é que isso é diferente com a IA?Karen Hao: De certa forma, não é diferente. Existe um fenômeno parecido: a indústria de IA terceiriza muitos dos trabalhos que ela não quer dentro dos centros de poder, e joga isso pra comunidades empobrecidas, do mesmo jeito que outras multinacionais fizeram por décadas.Mas também é diferente, porque a escala dos impactos trabalhistas e ambientais da IA é completamente outra, muito maior que a da indústria automobilística ou da moda. E a velocidade é outra, porque são tecnologias digitais que atravessam fronteiras muito rápido.E é diferente porque a maioria das pessoas não percebe que a IA, mesmo sendo tecnologia digital, tem uma cadeia de suprimentos muito física e intensiva em mão de obra manual.Quando você compra roupa, café, um carro, é mais óbvio que existem materiais que precisam ser extraídos e depois manuseados por pessoas pra criar aquele produto. Já com a IA, a maioria aceita a narrativa que o Vale do Silício projeta: a de que isso vem da “nuvem”, desses espaços etéreos que parecem nem existir no planeta. E a verdade é exatamente o oposto.Ela depende de uma quantidade extraordinária de extração mineral. Depende da construção de infraestruturas enormes — data centers, instalações de supercomputação espalhadas pelo mundo. E depende de muita, muita mão de obra manual: trabalhadores de dados que limpam, preparam e moderam o conteúdo dos sistemas de IA que chegam até você quando usa o ChatGPT.É isso que a torna tão diferente. E há também uma ideologia completamente diferente sustentando a expansão da IA. Quando você conversa com executivos da moda, eles não vão dizer: se você não comprar nossa roupa, vai pro inferno.Já a indústria de IA diz: se você não nos deixar capturar cada vez mais terra, mais recursos e mais mão de obra pra produzir essas tecnologias, vamos ter uma destruição civilizacional. Isso é, ao mesmo tempo, retórica política usada como arma pra moldar o debate público e a cabeça de quem formula políticas, e também está enraizado num sistema de crenças — algumas pessoas dentro dessas empresas realmente acreditam que, se uma AGI fosse construída, e construída nas mãos erradas, isso levaria mesmo a esse tipo de destruição.E é isso que move a sede cada vez maior da indústria por mais capital, mais recursos e mais terra.Cris: Eu quero falar sobre AGI, mas antes: ano passado, a OpenAI estava sendo processada no Reino Unido por violação de direitos autorais, basicamente todos os livros do mundo digitalizados e usados pra treinar modelos. E um dos executivos disse ao júri: bem, se a gente não puder fazer isso, fecha as portas. Me chocou que muita gente reagiu com um “ah, tá, o que a gente pode fazer? Eles vão fechar as portas”.Em parte porque a gente já está acostumado com essa narrativa. Outro dia, numa conferência, um ex-CEO dizia: a gente teve que usar embalagem de plástico porque é mais barata que papel, senão prejudicaria nosso resultado. E a plateia reagia: ah, então é só fechar as portas — a sociedade não pode arcar com isso.Mas isso também, como você disse, se conecta à ideia de uma grande missão, uma missão de salvar o mundo, que a gente precisa cumprir antes que seja tarde, senão estamos condenados. OpenAI está literalmente no nome — só que em português não é tão direto: é “inteligência artificial aberta”.Foi criada a partir de um sonho, um projeto que era pra ser uma coisa pro bem comum. Em 2019, num tempo bem distante, antes da pandemia, você cobriu a OpenAI, foi até o escritório deles, ficou lá dentro. O que você viu? E, mais importante, como essa missão mudou? O Elon Musk os processou outro dia justamente por mudarem a missão. Isso alguma vez foi verdade? Em algum momento eles pensaram mesmo “ah, a gente vai salvar o mundo”?Como essa narrativa de ser aberta funciona com a OpenAI?Karen Hao: Quando comecei a cobrir a OpenAI, levei a sério o que eles diziam — que tinham sido recrutados com a missão de beneficiar toda a humanidade. E aí, quando me infiltrei na empresa, fui ficando bem mais cética, porque via como eles operavam de um jeito completamente diferente, portas adentro, do que diziam em público.Diziam que iam publicar todas as pesquisas e abrir o código de tudo, e na prática eram uma das organizações mais secretas que já cobri. Eram muitas discrepâncias, e, na época, presumi que tinha havido algum tipo de corrupção que os levou a abandonar a missão original. Depois de cobrir a empresa por mais alguns anos e de trabalhar neste livro, mudei de ideia até sobre a missão original.Não acho mais que ela era um esforço sincero e generoso de beneficiar a humanidade. A missão foi criada pra dar à empresa — na época, uma organização sem fins lucrativos — uma margem de manobra extraordinária pra depois levantar muito capital, acumular muito talento e perseguir a força motriz de verdade por trás de tudo aquilo: se tornar a força dominante no desenvolvimento de IA.E penso assim agora porque, quando você olha pras narrativas de cada nova empresa de IA no começo — a Anthropic, a xAI, a Safe Superintelligence do Ilya Sutskever, a Thinking Machines Lab da Mira Murati —, todas usam a mesma narrativa da OpenAI: nós somos os mocinhos, eles são os bandidos.É por isso que precisamos criar uma nova empresa que avance a IA do nosso jeito, não do deles. E você começa a perceber, por esse padrão, que eles repetem a mesma coisa em parte porque acreditam nela até certo ponto, mas também porque ela funciona muito bem com a imprensa, com o público, com quem formula políticas.No livro, eu reproduzo os e-mails internos que Elon Musk, Sam Altman e Greg Brockman trocavam nos primeiros dias da OpenAI. Eles tinham plena consciência de que estavam criando uma missão que soasse bem para o público. E o propósito de verdade, que também deixaram registrado nesses e-mails, era vencer o Google. Viam o Google como a força dominante em IA e queriam ser eles essa força.Não gostavam de ver o Google na frente, então inventaram justificativas: o Google é uma empresa com fins lucrativos, então nós vamos ser sem fins lucrativos. Mas, no fundo, acho que era puro ego: tem que ser a gente, não eles, a gente quer ser quem lidera isso.E aí passaram um tempão moldando essa missão pública, que acabou sendo super útil pra recrutar o primeiro grupo de pesquisadores e turbinar o avanço deles.Cris: Então agora é um bom momento pra falar de AGI, a inteligência artificial geral. Muita gente pergunta: o que é AGI? O que “geral” quer dizer? E a impressão que peguei lendo seu livro é que, por design, isso nunca fica claro de verdade, porque é um alvo móvel. Essas empresas um dia vão dizer “chegamos, alcançamos a AGI”? Ou o plano é sempre “não, não, ainda não chegamos, me dá mais dinheiro, me dá mais poder”?Qual é o papel da AGI na narrativa dessas empresas?Karen Hao: Já que a gente está falando de ficção científica, eu costumo usar a analogia de que o mundo da IA é meio como Duna. Em Duna, o personagem principal, Paul Atreides, entende, ao chegar no planeta Arrakis, que o povo de lá foi semeado com um mito: o de que um dia viria um Messias pra libertá-los. Ele sabe que é um mito, mas decide entrar nele e agir como se fosse o Messias pra controlar melhor aquele povo.E, vivendo, respirando e encarnando esse mito dia após dia, ele começa a perder a noção de que é um mito. Passa a se perguntar se o mito era mesmo verdadeiro ou se foi ele quem o tornou verdadeiro. É essa confusão entre mito e realidade — ele vive num espaço intermediário, sem ter mais certeza do que é verdade e do que é ficção.E trago isso pra responder sobre a AGI porque a AGI é, ao mesmo tempo, um mito e algo que os líderes e os trabalhadores dessas empresas vivem, respiram e encarnam dia após dia, a ponto de perderem a noção do que é mito e do que é realidade. É a ideia de um sistema de IA teórico que um dia igualaria as capacidades humanas. Só que a gente nem tem consenso científico sobre o que é inteligência humana.Por isso, de certa forma, por design, é um termo bem maleável, que deixa essas empresas fazerem o que quiserem. Elas definem e redefinem a AGI conforme a necessidade, movem a trave pra onde quiserem. E, ao mesmo tempo, isso é sustentado por uma crença genuína de certas pessoas lá dentro, por causa dessa confusão entre mito e realidade. Pelas minhas contas, a OpenAI já usou pelo menos quatro definições diferentes de AGI.A primeira está no site deles: “sistemas altamente autônomos que superam humanos na maioria dos trabalhos economicamente valiosos”. É uma definição de automação do trabalho — eles dizem, de forma explícita, que estão atrás dos empregos mais bem pagos. A segunda apareceu no contrato com a Microsoft, por um tempo a maior investidora deles: ali, a AGI virou um sistema que geraria 100 bilhões de dólares em receita.Ou seja, uma definição feita pra incentivar a Microsoft a investir. Já o Sam Altman disse ao Congresso que AGI é um sistema que cura o câncer e resolve a mudança climática — uma definição de benefício social, muito útil quando você quer que os reguladores não te regulem.E, por fim, quando falam com o consumidor, dizem que vai ser o melhor assistente digital que você já teve — porque, claro, estão tentando vender o produto.E aí você percebe duas coisas. Primeiro, que é um conjunto de definições completamente incoerente. Segundo, que eles trocam de definição conforme o público que querem convencer. Mas também tem gente nessas empresas que acredita de verdade que está construindo uma tecnologia capaz de dar conta das quatro coisas.Então é uma realidade bem confusa e complicada: o que a AGI de fato é, e pra que ela serve, para essas empresas, para a agenda delas e também para as crenças delas.Cris: Como ex-funcionário da Meta — entrei em 2013 —, a missão era unir o mundo e torná-lo mais aberto e conectado. É uma missão incrível. E tem uma coisa que eu sempre digo, porque muito amigo meu vem falar comigo, “ah, esse cara da OpenAI, ou a própria Meta, são maus”. Eu conheci muita gente na empresa. Nunca conheci uma pessoa mal-intencionada.Todo mundo, independente da missão, era gente boa tentando entregar o melhor produto possível, pra dar poder a quem tem um pequeno negócio, por exemplo. Tenho amigos pessoais que construíram a empresa deles em cima da publicidade do Facebook e do Instagram. E esse é justamente o problema, porque ainda assim é uma corporação muito má, pelo que ela causa ao mundo pra bater as metas de negócio.Ou seja, você não precisa de um vilão tipo Lex Luthor pra causar um estrago desse tamanho no mundo. E adorei a referência a Duna. Duna é engraçado: é o livro que eu mais reli na vida que não foi escrito pelo Tolkien. Li o primeiro Duna umas três vezes, e toda vez é como se fosse um livro diferente. Na primeira, eu era adolescente, e era só o Paul Atreides, o cara durão.Na segunda, eu morava no Canadá e li com olhos de estrangeiro, pensando em colonização. E na terceira vez foi quando os filmes do Denis Villeneuve saíram, e aí era: ah, o Bene Gesserit criou esse mito, isso é meio pós-moderno. Narrativamente, fico me perguntando o que vai significar pra mim se eu ler uma quarta vez.Karen Hao: Eu ia te perguntar isso. Quando você disse que cresceu numa época cheia de ficção científica falando das maravilhas da tecnologia, fiquei curiosa: que histórias você estava lendo? Porque muita coisa que saiu nos anos 70 e 80 dizia exatamente o oposto. E muita gente já apontou que os executivos de tecnologia de hoje, que vivem citando essas histórias, interpretam elas justamente ao contrário da intenção original.Cris: Concordo plenamente. Mas, respondendo: foi basicamente Isaac Asimov e Arthur C. Clarke. E é por isso mesmo — os executivos de tecnologia, e o Elon Musk mais que todos, leem esses livros como receita, não como aviso. O livro de que eu mais me lembro, nem lembro o título, era um do Asimov em que ele descreve o elevador espacial que aparece na série da Apple TV, Fundação.E o enredo é: eu sou esse engenheiro brilhante, quero construir essa coisa no Sri Lanka, mas o governo trava tudo com regulação — eu sou um gênio e a regulação é a vilã. Hoje eu leio e penso: ah, sei. Mas, quando garoto, era só “olha, um elevador espacial, que genial, a gente nem precisa de foguete”. E aí você começa a entender. E aí eu parei de ler esses caras.E passei a ler gente com uma visão completamente diferente: o Ted Chiang, que entrevistei ano passado, a N.K. Jemisin, o Cory Doctorow, de quem sou muito fã. E talvez eles sejam mais explícitos, pra gente burra como eu entender: “não, bobo, a analogia é essa”. Mas Duna era incrível — vermes gigantes de areia, o tal garoto durão, e aquela coisa do “eu não aceito o meu destino”.Tenho esse grande destino, mas não quero ele. Do resto da série eu já não gosto tanto. Mas o mais importante de tudo: Duna gerou o melhor GIF de filme de todos os tempos, o “Lisan al Gaib” do Javier Bardem — que eu devia ter colocado durante a sua explicação, aquele “uau, ele está cumprindo a profecia, agindo como o profeta”.Mas, de novo, falando de vilões: você mencionou que essas empresas se colocam como o bem contra o mal, feito impérios antigos. Só que elas também jogam a carta da China, né? “Se a gente não fizer, a Rússia faz primeiro.” Só que a Rússia começou uma guerra e está ocupada demais. “Mas a China chega lá, e é por isso que a gente tem que ser fechado.” É por isso que Mythos e Fable e agora o GPT-5.6 foram proibidos pelo governo. Isso tem fundamento?Quero saber se é possível a China competir — quero mesmo essa resposta — mas também porque, desde toda essa conversa do Fable-Mythos, países como Índia e Brasil vêm dizendo que precisam de um modelo soberano. Dá pra fazer, ou a OpenAI, a Anthropic e o Google estão tão à frente que já não dá?Karen Hao: Sobre a China: você está certíssimo, o Vale do Silício usou por anos a carta do “e a China?” pra escapar de qualquer responsabilização de verdade. Fizeram muito isso na era das redes sociais.A Meta fez muito isso, com o Mark Zuckerberg dizendo ao governo dos EUA: vocês não podem nos regular, senão a gente perde. Mas, se a gente ganhar, vai ter um efeito liberalizante no mundo e nas democracias em todo lugar. E, infelizmente, o que a gente viu foi que jogar essa carta repetidamente produziu exatamente o efeito contrário do que o Vale do Silício prometeu.Uma das empresas de rede social dominantes dessa era é a ByteDance. Ou seja, mesmo sem regulação das redes sociais nos EUA, existe uma empresa chinesa de rede social bem dominante. E as redes sociais estadunidenses acabaram tendo um efeito antiliberal no mundo — é bastante consensual que enfraqueceram democracias em todo lugar. E aí, na era da IA, elas seguiram jogando a mesma carta.Mas o que eu sempre aponto é que a gente definitivamente não devia acreditar nelas. Já existe evidência significativa de que tudo o que elas dizem está, de novo, se provando o oposto. Elas disseram: não regulem a gente como empresas de IA, regulem a China, via controles de exportação — um mecanismo do governo dos EUA com alcance extraterritorial.Só que as empresas chinesas agora estão produzindo modelos de IA de código aberto extremamente eficientes, que viraram super populares no próprio Vale do Silício. Existe um monte de startup de lá que prefere usar modelo chinês a OpenAI, Anthropic ou Google.Então, nesse sentido, é um conjunto de evidências bem decisivo, acho, pra mostrar que a gente devia simplesmente responsabilizar essas empresas, não importa o que digam sobre “ah, vamos perder pra China”. No fim das contas, é só retórica política. Não é um argumento real que elas consigam sustentar pra escapar da responsabilização.Responsabilizá-las vai fortalecer a democracia pelo mundo, vai trazer mais direitos humanos, trabalhistas e de privacidade de dados pras pessoas — é sempre o contrário do que elas dizem que aconteceria. E, sobre a sua pergunta em torno da IA soberana: acho a ideia realmente importante, mas acho também que muitos países estão meio confusos sobre o que querem dizer com isso.Muitos governos, hoje, pensam a IA soberana pela pergunta: a gente consegue construir o nosso próprio ChatGPT? O nosso próprio grande modelo de linguagem, o nosso sistema de IA generativa? Estão olhando só pro modelo que o Vale do Silício já definiu e tentando descobrir como recriar aquilo.E o que eu digo pra quem formula políticas é: defina pra que a IA serve no seu país, no seu contexto. Quais são, no fim das contas, os objetivos do seu país? Os objetivos do seu povo? E também os nossos objetivos coletivos, entre países?Porque a gente tem, por exemplo, os Objetivos de Desenvolvimento Sustentável da ONU. Já definimos coletivamente que há coisas que precisamos resolver juntos: superar a crise climática, reduzir a pobreza, melhorar a educação.E, enquanto o Vale do Silício adora dizer que está fazendo tudo isso, na prática não está. Mas a gente poderia — poderia desenvolver, de forma colaborativa, sistemas de IA que realmente avançassem em cada um desses objetivos coletivos que já acordamos.E cada país também devia fazer o exercício: quais objetivos você quer alcançar, e que tipos de sistema de IA você poderia desenhar pra chegar lá — sistemas que talvez não se pareçam em nada com um grande modelo de linguagem. Se os países fizessem isso, acho que descobririam que a maioria dos sistemas de que precisam exigiria muito menos recursos.Ou seja, contextos como o Brasil, a Índia e outros, quando não precisam competir construindo essas infraestruturas de computação gigantescas e gastando centenas de bilhões de dólares, na verdade já têm, localmente, todos os recursos necessários pra desenvolver um sistema de IA soberano.Cris: Quando ouvi falar do seu livro pela primeira vez, uma amiga me disse que você não poupa ninguém — fala mal do Sam Altman, mas também do Dario Amodei. E as pessoas costumam escolher um lado. Eu sou time Claude, odeio o ChatGPT, essas coisas. Então cheguei no livro pensando: ah, é mais um livro dizendo que a IA é terrível, que a gente não devia usar IA.Mas, conforme fui lendo, e ouvindo outras entrevistas suas, me pareceu que o seu problema é justamente o que você acabou de descrever: a forma como essa tecnologia é feita. E, em especial, a palavra escala — a ideia de que a solução é a escala. O que você quer dizer com isso?Karen Hao: Eu costumo usar a analogia de que “IA” é como a palavra “transporte”: na verdade se refere a uma coleção de tecnologias que vão da bicicleta ao foguete. São tipos bem, bem diferentes de tecnologia, que exigem insumos diferentes pra se desenvolver e depois têm impactos diferentes na sociedade.E você está certo: sou especificamente crítica ao que chamo de “foguetes da IA”, os sistemas que os impérios da IA estão desenvolvendo, aqueles que exigem uma quantidade enorme de exploração de mão de obra e extração ambiental.E sou bem otimista com o que chamo de “bicicletas da IA”: sistemas especializados, eficientes, com bom custo-benefício, governáveis pelas pessoas, cujo desenvolvimento pode ser participativo. Países como o Brasil, o Chile, a Índia, qualquer contexto, têm recursos pra desenvolver e se autodefinir, em vez de simplesmente herdar um sistema criado pelos dois únicos centros do mundo capazes de gastar uma quantidade extraordinária de capital: o Vale do Silício e o ecossistema tecnológico chinês.E o motivo pelo qual eu acho tão corrosivo o que os impérios da IA estão desenvolvendo é exatamente o que você disse: a forma como eles fazem isso, por um mecanismo de força bruta pra avançar as capacidades da IA em escala. Eles vão simplesmente empurrando cada vez mais dados de treinamento nesses modelos, e isso exige corroer a privacidade das pessoas, tomar a propriedade intelectual delas e, ainda por cima, baixa a qualidade dos dados que entram nos modelos — o que leva aos danos de exploração de mão de obra, porque aí você tem que dar conta da moderação de conteúdo.E aí você tem pessoas psicologicamente traumatizadas por serem expostas a todo aquele conteúdo horrível que se tenta “lavar” através desses modelos.E aí vem o problema dessas infraestruturas de computação enormes, com impactos ambientais que aumentam a conta de luz das comunidades que as hospedam e agravam a crise de custo de vida. Elas precisam ser alimentadas por fontes fósseis, que jogam mais carbono na atmosfera e mais poluição no ar dessas comunidades.Então todos os problemas que eu identifico, no que têm de corrosivo, derivam inteiramente da abordagem deles pro desenvolvimento de IA. Por que não descartar a abordagem, em vez de descartar a tecnologia? Redefinir e redesenhar de que tipos de sistema de IA a gente precisa de verdade, com uma cadeia de suprimentos fundamentalmente diferente. E isso não é exclusivo da IA.A gente já viu muitas outras indústrias que começaram com uma cadeia de suprimentos bem ruim. A moda, por exemplo: muita degradação ambiental, muita exploração de mão de obra.Com muita organização, protesto, ação de consumidores, regulação governamental e cooperação entre governos, a gente conseguiu criar mercados novos pra moda sustentável e ética, cadeias de suprimentos novas e inovações pra fazer roupa mais saudável pras pessoas e pro planeta.E é basicamente isso que eu defendo: transformar a indústria de IA do mesmo jeito que transformamos a moda, e as cadeias de suprimento de alimentos. Assim a gente fica com os benefícios da tecnologia, ajuda ela a avançar os objetivos que importam pra gente, sem jogar uma fração enorme da população mundial numa condição atrasada e numa qualidade de vida pior.Cris: O Brasil está agora, no Congresso, discutindo a escala de seis dias por semana. A regra atual é: você trabalha seis dias e descansa um. E muitas empresas, o comércio principalmente, dizem “vamos fechar as portas”, e os trabalhadores respondem “isso é problema seu, não meu”. É mais ou menos a mesma narrativa dessas empresas de IA: se eu não usar a sua água, a Idade das Trevas está chegando.Falando em Idade das Trevas, e falando em bicicleta: a sua analogia me lembrou uma coisa. Eu gosto de jogo de zumbi, de mundo aberto, e em nenhum deles tem bicicleta. Num desses jogos, instalei um plugin que deixava andar de bicicleta — você acha uma e sai pedalando. E aí entendi por que não tem bicicleta: desbalanceia tudo. Parte da graça do jogo é você precisar achar um carro, e daí pneu, gasolina, comida pra carregar. De bicicleta, você vai a qualquer lugar.E eu pensei: ah, é. Meio que estraguei o jogo pra mim, porque agora tenho uma bicicleta, é incrível. Enfim, em termos práticos: no fim do ano passado, uns meses atrás, a revista Wired publicou um artigo pedindo pra jornalistas de tecnologia contarem como usam IA no trabalho. E cada um usava de um jeito. Você usa IA no seu trabalho? Como?Karen Hao: Eu não uso nenhum sistema de IA generativa no trabalho — nem ChatGPT, nem Gemini, nem Claude. Por três motivos. O primeiro é uma postura ética, depois de tanto investigar essas empresas. O segundo é privacidade de dados: eu investigo essas empresas.Não quero que elas conheçam todo o meu raciocínio enquanto eu apuro o livro, literalmente investigando elas. E o terceiro é que, no meu caso específico, a força do meu trabalho está na capacidade de construir relações fortes com as fontes, pela empatia, e de contar histórias envolventes, pela narrativa. E os grandes modelos de linguagem simplesmente não são a ferramenta certa pra nenhuma das duas coisas.Não vão melhorar a minha empatia nem a minha escrita. Então eu não perco nada com essa postura ética: simplesmente corto essas ferramentas e sigo fazendo o meu trabalho muito bem. Pra outros jornalistas pode ser diferente, e pra quem está em outras áreas o cálculo pode ser outro.Mas eu incentivo as pessoas a pensarem primeiro: quais são as suas forças no trabalho? Quais são os seus objetivos? E aí ir de trás pra frente pra descobrir se a IA é a ferramenta certa, qual tipo de IA é a ferramenta certa, e qual fornecedor você quer de fato usar, apoiar, votar com os pés. Agora, eu uso, sim, IA preditiva.Aquelas ferramentas de IA especializadas, as “bicicletas da IA”, digamos. No livro, tinha um detalhe que eu queria muito ilustrar: como a OpenAI deu um salto quando passou de organização sem fins lucrativos a um empreendimento bancado pela Microsoft. Percebi que as cadeiras do escritório ficaram bem mais caras. Então fotografei as cadeiras de um escritório e as do outro.E joguei tudo na busca reversa de imagens do Google, que é um sistema de IA especializado — não é baseado em grandes modelos de linguagem, não é IA generativa. Assim descobri quanto essas cadeiras costumam custar. No primeiro escritório, cerca de 2 mil dólares por cadeira. No segundo, eram cadeiras de um designer brasileiro famoso, uns 10 mil dólares cada.Coloquei esse detalhe no livro pra ilustrar o tipo de riqueza e de concentração de recursos de que a gente está falando. Esses são alguns dos jeitos como eu uso IA, ainda que de forma bem limitada, sempre pontual, quando acho que vai ajudar. E, claro, uso ferramentas de transcrição por IA — outra IA especializada — em todas as minhas entrevistas.Cris: Essa foi uma das partes em que a minha cabeça explodiu, eu nunca tinha percebido: a OpenAI criou o Whisper. Deixa eu dizer de outro jeito, do meu ponto de vista. A OpenAI liberou abertamente essa ferramenta incrível de transcrição, o Whisper, em que eu jogo o áudio e ela me devolve as palavras que as pessoas disseram. E eu pensei: ah, que generoso da parte deles.Mas o motivo real de terem criado a ferramenta foi pegar todos os vídeos do YouTube, transcrever e alimentar a máquina. E aí é: ah, claro. Enfim, falando de ferramentas e de otimismo — a gente está chegando ao fim da conversa. Eu tenho uma regra desde o episódio dois deste programa, há oito anos: de novo, como eu disse do seu livro, não pode ser só uma lista de reclamações e coisa ruim. E a gente tem se saído bem até aqui.Você falou de caminhos e de bicicletas, mas eu quero ser mais específico. Se isso aqui fosse uma reunião de negócios: qual é o plano de ação, quais são os próximos passos? Só que uma das coisas que eu repito bastante, na vida e neste programa, é que problema sistêmico não se resolve com ação individual. Se eu tomar banhos mais curtos, isso nunca vai salvar o planeta do aquecimento global.E muitos amigos meus simplesmente: não quero falar de IA, não quero usar IA. Voltando aos videogames: leram que tal jogo usa IA e pronto, não vão jogar. E a minha primeira pergunta pra você é: como a gente ocupa esses espaços da IA generativa — ChatGPT, Gemini e por aí vai? Porque o que a gente viu com as redes sociais foi: ah, o Facebook é do mal, vou sair do Facebook. Ah, vou sair do Twitter.E, na esperança de quê, sei lá, talvez alguém diga: ah, sinto falta do Cris, cadê ele? Ah, está no Bluesky. Mas isso deixa o espaço aberto pra os radicais entrarem e postarem o que quiserem, sem ninguém contrapor ou tornar aquilo um lugar melhor. Então como a gente ocupa o espaço da IA — seja qual for a definição de “espaço da IA” que você preferir — com todos esses problemas que a gente vem discutindo?Karen Hao: Acho que tem duas categorias de ação pra gente pensar. Uma é desmantelar o império. A outra é investir e construir novos tipos de sistema de IA, que se tornem alternativas às tecnologias do império. Quando eu digo desmantelar o império, não estou dizendo que quero que a OpenAI, o Google, a Anthropic, seja quem for, simplesmente deixem de existir.É que eu não quero que elas sejam imperiais. Não quero que fiquem extraindo uma quantidade extraordinária de valor sem redistribuir nada em troca. Se elas voltassem a ser negócios que praticam uma troca justa de valor com o mundo, eu ficaria perfeitamente feliz com qualquer tecnologia que estivessem desenvolvendo.E a forma de desmantelar o império, acho, se resume a muita organização de base, que vai pressionar os governos a regular e responsabilizar essa indústria. No último ano, a gente viu uma quantidade incrível dessa organização de base florescendo pelo mundo.Recentemente, lancei com um grupo de jornalistas, pesquisadores de IA e acadêmicos críticos um projeto chamado AI Resist List, que busca documentar parte dessa organização de base pelo mundo. A gente encontrou cerca de 30 exemplos, de todas as regiões, de ações individuais, institucionais e movidas pela comunidade.Tinha ação artística, ação política. E isso mostra bem o seu ponto: não dá pra contar só com a ação individual, mas o indivíduo pode, sim, ter impacto. Até uma ação pequena pode gerar um grande efeito cascata. Claro que se juntar com os vizinhos pra protestar contra o data center é ainda mais eficaz. Se juntar dentro da sua escola ou universidade pra protestar contra a parceria dela com uma empresa de IA também é mais eficaz.Se juntar com os colegas de trabalho de um setor pra barrar a adoção de uma IA que corrói os direitos trabalhistas é mais um jeito eficaz. A gente tem um monte desses exemplos. Um dos meus favoritos é o de uma comunidade sobre a qual escrevi no livro, Quilicura, no Chile, na periferia de Santiago. É uma comunidade da classe trabalhadora, bem pobre, que vem sendo alvo incessante da expansão de data centers.E por isso protestaram de forma bem aguerrida contra essa expansão, porque não acharam bom negócio hospedar essas instalações sem tirar nenhum benefício, enquanto elas consomem uma parte significativa dos recursos naturais da região.E, logo depois que escrevi sobre eles, foram além na resistência e criaram uma plataforma chamada Quili.ai. É um site em que você entra e que parece um chatbot, parece o ChatGPT: tem uma interface de chat pra você digitar. Só que, quando você faz uma pergunta, em vez de um modelo de IA responder, a mensagem é encaminhada pra alguém que mora em Quilicura, no Chile. Aí, se você pede “quero a imagem de um cachorro”, aquilo vai pro artista local deles, o Benji. Ele pega um pedaço de papel, desenha um cachorro, tira uma foto e te manda de volta.Eles fizeram isso essencialmente como um projeto de arte performática, pra fazer as pessoas pensarem duas vezes antes de usar IA generativa pra bobagem. A mensagem era: ei, quando você fica brincando com essas ferramentas em pedido besta, isso afeta comunidades como a nossa, drena os recursos de que a gente precisa pra viver bem.E também queriam levar as pessoas a pensar: por que não perguntar pra alguém da sua própria comunidade aquela receita que você procurava, ou pedir aquela imagem? Porque aí você reconstrói as conexões que estão tão em falta na sociedade — a falta delas é o que nos deixa mais vulneráveis a esse tipo de colonização do império.Eles deixaram o projeto aberto por 24 horas, e qualquer pessoa no mundo podia mandar um pedido. Receberam uma quantidade extraordinária deles. Viralizou de vez. E essa cidadezinha conseguiu uma virada enorme de narrativa sobre a suposta inevitabilidade e necessidade dessa tecnologia, sobre tudo o que o Vale do Silício diz — que, se você não usar, vai ficar pra trás de quem usa.E esse é só um exemplo, entre muitos, de como pessoas comuns, não importa a sua posição na sociedade, podem ter impacto real no debate, na consciência pública e até na regulação. A gente está vendo isso agora com os protestos contra data centers. Nos EUA, em 2025, cerca de 150 bilhões de dólares em projetos de data center foram travados.Isso virou uma das questões políticas mais quentes nos EUA para as próximas eleições de meio de mandato. Tem gente eleita sendo literalmente tirada do cargo por ter aprovado data centers, contrariando a vontade do povo. E isso já está tendo efeito real sobre as empresas e sobre a trajetória do desenvolvimento de IA.A OpenAI teve que encerrar recentemente a sua ferramenta de geração de vídeo, o Sora. Quando lançaram, apresentaram como o segundo produto mais importante desde o ChatGPT. O que aconteceu entre o lançamento e o fim? Uma reportagem do Wall Street Journal apontou três motivos, todos moldados por ação de base. Um: um gargalo enorme de capacidade de computação.Muitos dos data centers travados ou parados eram da OpenAI. Dois: um cenário financeiro bem mais incerto. A OpenAI está se preparando pro IPO, o que significa ficar mais exposta a Wall Street — e Wall Street está cada vez mais nervoso com a capacidade dessas empresas de cumprir o que prometem.E aí a OpenAI teve que reforçar alguns projetos paralelos pra fazer o balanço parecer um pouco melhor aos olhos de Wall Street. E, terceiro: os consumidores simplesmente não estavam usando o produto — o que também é ação coletiva de consumidores. Então, por todo esse tipo de resistência, de várias formas, de baixo pra cima, as pessoas estão de fato tendo impacto real na indústria e responsabilizando ela.Essa é a primeira categoria de ação. A segunda é: ok, que tecnologias de IA a gente usaria como alternativa? E aí a gente precisa investir mais nelas. Muitas vezes, quando converso sobre o livro, a pessoa diz: ok, me convenci de que não quero usar ChatGPT, não quero usar Claude — mas então uso o quê no lugar?E o problema é que eu não tenho muitas respostas pra essa lista de alternativas. Tem umas poucas aqui e ali, uma plataforma, uma empresa.Cris: Dá pra rodar o modelo no seu próprio computador, como o Cory Doctorow faz, mas aí é limitado e…Karen Hao: Exatamente, exige mais habilidade técnica. Mas, pra quem consegue instalar modelos de código aberto no próprio computador, eu incentivo 100%. Só que a gente também precisa de mais gente desenvolvendo interfaces bem fáceis pra esses modelos de código aberto, pra que qualquer pessoa consiga usar.A gente também precisa de mais gente desenvolvendo “bicicletas da IA”, de investidores e governos investindo mais nesse tipo de solução, e de talento — pesquisadores de IA, desenvolvedores e outras pessoas dispostas a sacrificar um pouco e abrir mão dos pacotes de remuneração enormes.Cris: Eu estava começando a achar que agora as empresas precisam ter menos lucro — e isso nunca vai acontecer.Karen Hao: Não, não é a empresa ter menos lucro. É o trabalhador topar abrir mão do pacote de milhões de dólares pra levar o talento dele pra outro lugar. Mais fácil, bem mais fácil. Eu converso com muito pesquisador de IA cansado da abordagem da indústria, porque ela é completamente sem criatividade intelectual.Eu conversei com pesquisadores que não passaram seis anos num doutorado em IA só pra ficar empurrando mais dados na máquina — pra eles, é o trabalho mais chato do mundo. E depois automatizar a programação, que era justamente o que eles gostavam de fazer. Converso com tanta gente que já não acha graça nenhuma nisso. Estão meio presos por “algemas de ouro”.E estão tentando descobrir, dentro de si, que carreira alternativa poderiam ter. Eu costumo incentivar esses pesquisadores a gastar o talento deles construindo um tipo diferente de empresa, que trabalhe com “bicicletas da IA”. E a gente já começa a ver cada vez mais desse talento indo por aí.E a gente precisa que todas as facetas da sociedade invistam num ecossistema muito mais robusto e rico de tecnologias de IA, capaz de substituir as que hoje dominam. Eu ainda tenho as cicatrizes das minhas próprias “algemas de ouro”, mas concordo plenamente.Cris: E as redes sociais são o exemplo — veja o que aconteceu com elas. Tem aquela frase famosa: as mentes mais brilhantes da minha geração passam o tempo fazendo as pessoas clicarem em anúncios. E ainda dizem: ah, isso pode ser o futuro. Pois é.Você contou a história do Quili.ai e isso me lembrou um dos primeiros criadores de conteúdo do Brasil, o Cid Não Salvo. Uns 10, 15 anos atrás, ele tuitou o seguinte: “Gente, eu disse pro meu pai que, sempre que ele precisar pesquisar alguma coisa na internet, é pra ir no Twitter.com e digitar a pergunta na caixa”. E olha que ele tinha milhões de seguidores.E, por uns bons dias, quase um mês, você entrava no Twitter do pai dele e via perguntas tipo “onde eu compro pizza?”. Era engraçadíssimo. No fim, ele contou pro pai — ou talvez não. Mas eu adoro essa ideia. Antes de a gente terminar: você já deve ter respondido isso mil vezes, mas vai continuar cobrindo IA? O que está na sua cabeça, o que vem por aí? Turnê mundial? O que vem pela frente?Karen Hao: Com certeza estou pensando em como continuar responsabilizando essas empresas. Estou envolvida em várias colaborações, com gente incrível, em diferentes projetos ligados a isso. O AI Resist List foi um deles. Também co-criei um programa chamado AI Spotlight Series, com o Pulitzer Center, uma organização jornalística sem fins lucrativos que financia jornalismo investigativo pelo mundo.É um programa que treina jornalistas do mundo inteiro a cobrir IA por uma lente de responsabilização. Até agora, já treinamos mais de 3 mil. E eu sigo pensando em como construir mais capacidade dentro do jornalismo, da sociedade civil, de outros contextos, pra mobilizar ainda mais essa organização de base — pra conter de verdade os impérios da IA e ajudar a desmantelá-los.Cris: Adorei o seu exemplo da moda. É possível, já foi feito. Ou até a indústria automotiva. Ou o grande exemplo que a gente não mencionou, e que o pessoal da OpenAI vive citando: o Projeto Manhattan, a energia nuclear.O mundo não acabou. Quando eu era criança lendo Asimov, achava que ia tudo acabar num fogo nuclear. Enfim — alguma última palavra, alguma mensagem, algum palpite pros jogos do Brasil na Copa, alguma coisa que você queira dizer antes da gente encerrar?Karen Hao: No fim das contas, o que eu espero que fique desta conversa e do livro é o seguinte: neste momento, o Vale do Silício está concebendo a IA como um projeto político. E a característica central desse projeto é tirar a autonomia de todo mundo — a autonomia de moldar de verdade o próprio futuro e o nosso futuro coletivo. Mas, no instante em que você reconhece que já tem uma autonomia significativa pra resistir, o império começa a desmoronar.Então espero que as pessoas encontrem a própria voz, a afirmem, conquistem o seu lugar à mesa e se conectem com os vizinhos, com a comunidade, com os colegas de trabalho, pra criar mais movimentos juntos.Cris: Que ótimo. Karen Hao, o seu livro é O Império da IA: Por dentro da corrida irresponsável pela dominação total. Obrigado por vir ao Brasil conversar com a gente. Foi um prazer.Karen Hao: Muito obrigada.Uma das primeiras perguntas que anotei quando comecei a pensar nessa conversa foi justamente a do final, a da ocupação de espaços. Porque, como eu disse, quando as redes sociais chegaram para ficar, muita gente falou “ah, não vou usar, é do mal” — e aí as pessoas ruins, vamos chamar assim, acabam ocupando esse espaço e falando o que bem entendem. A gente precisa aprender essa lição agora, no mundo da IA.Fora que vejo muita gente falando de IA sem nunca ter usado — ou que usou, sei lá, dois anos atrás, acha que continua tudo igual e já diz que não quer chegar perto.E por quê? Porque essa abordagem de ocupar espaços é o que eu e a Ana Freitas buscamos fazer no IA em Curso, nossa comunidade de letramento contínuo em IA. Foi, aliás, uma conversa que tive com a Karen antes da entrevista: ao mesmo tempo que a gente fala do impacto da IA no mundo, também precisa focar no que é prático, no que dá para fazer hoje com IA, sem vender sonho nem desastre. A analogia que usei foi a de que é que nem quando a gente fazia curso de Word e Excel — é o que eu faço agora que vai facilitar minha vida, me fazer ganhar tempo, botar a IA para me ajudar. Quem viu minha conversa com a Ana aqui no Boa Noite Internet, no fim de 2025, sabe do que estou falando. Se não viu, volta lá e confere.Desde que a gente lançou este episódio, o IA em Curso já passou de 400 pessoas. Tem muita gente colocando projetos pessoais incríveis na rua, tirando do papel aquela ideia que rondava a cabeça há um tempão. E a comunidade tem mentoria ao vivo, aula gravada, newsletter, banco de agentes, grupo de Telegram… que mais? O que não falta é jeito de passar para você o conhecimento sobre IA de que você precisa hoje, agora. Quero te dar a bússola para navegar nesse universo.Se esse é o tipo de abordagem que você quer ter com a IA, passa lá no iaemcurso.com.br e usa o cupom BNI2026 para ganhar 20% de desconto no plano anual. Mas corre, porque daqui a duas semanas vou apagar esse cupom — não é todo dia que a gente dá um desconto desses.É isso. Boa Noite Internet, temporada 2026 começando — como todo ano, com mudança, ideia, projeto. Ou, como diz minha citação preferida de todos os tempos: “vivemos uma fase de transição, como sempre”. Espero ver você por aqui e lá no IA em Curso.Obrigado pelo seu tempo e pela sua atenção. Até o próximo episódio. This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit boanoiteinternet.com.br/subscribe

Polymarket isn't just a betting app — it might be the newest branch of the surveillance state. This week on Conspiracy Social Club, we break down journalist Whitney Webb's bombshell investigation tying Polymarket's origins to Palantir, Peter Thiel, DARPA's Information Awareness Office, and the Pentagon's decades-long obsession with predictive markets. We trace the lineage from John Poindexter's Iran-Contra-linked "Policy Analysis Market" all the way to today's prediction-market boom — plus the Zuckerberg and Bancor connections nobody's talking about. Then things get heated: Trump's claims of a massive Chinese data breach exposing 220 million voter profiles kicks off a full-blown 2020 election fight — FISA warrants, Carter Page, Thomas Massie's "Deep Throat" theory, election recounts in Arizona and Georgia, mail-in ballot security, and whether democracy itself is already broken. As always, it's not a debate without chaos — comment of the week, TRT talk, and way too many tangents. Get in the comments and let us know where you stand.

In 1776 — that same year America declared its independence — Adam Smith published the equally revolutionary The Wealth of Nations, his founding explanation of national economic value. Two hundred and fifty years later, Tim O'Reilly argues in the free-market Economist that Elon Musk and his fellow tech barons are building a monarchical form of capitalism that the proto-democratic Smith would have hated. Musk, O'Reilly reports, believes that SpaceX will become “worth more than the rest of Earth”. The merchants are becoming princes, O'Reilly warns. And the rest of us are becoming peasants. Such is the road to serfdom in our AI age. So who should own the AI in our bewildering age of multi-trillion dollar start-ups like SpaceX, Anthropic and OpenAI? Or as That Was The Week publisher Keith Teare asks in his latest editorial, who should own the “intelligence” of our AI age? Keith uses a bottling plant as a metaphor to describe our dilemma. Since no single entity can own this intelligence — the sum total of our common experience — charging us for it would be like seizing the Earth's water supply and selling it back to us, Coca-Cola style, in plastic bottles. Except that the Hayekian Keith approves of the bottling process. Private companies, rather than governments, he argues, are most suited to doing this. For Keith, this dilemma is also an opportunity to redistribute the ownership of intelligence. He argues for a “Human Wealth Fund” into which every consequential AI company should put a slice of its equity. In the manner of Norway's sovereign wealth fund, this fund would be distributed to all citizens. Rather than Denmark, now we should become like Norway, a tiny homogenous nation with a cultural distaste for Muskian individual wealth. Not very realistic, I fear. On top of that, it's hard to imagine our tech princes collaborating on anything. Musk and Altman aren't on speaking terms while Altman and Amodei, who also loathe each other, are focused on their IPOs. Meanwhile, the Trump administration, which presumably would coordinate this fund, is pitching a $100,000-a-month fast feed of the president's posts. Keith's question, “who owns the intelligence”, is the right one. But the answer won't come from trickle-down funds set-up by our tech princes. Such supposed munificence is about as likely as America becoming Norway. Read the fine print of any “Human Wealth Fund” set up by Sam Altman and Elon Musk. As we should know all too well by now, when a “revolutionary” Silicon Valley gives stuff away, it turns out to be exorbitantly expensive. Free plastic bottles of intelligence, anyone? Five Takeaways •       Intelligence, Not AI. The week's framing shift: the word AI is too small, because AI is merely the tool for harvesting and delivering the thing itself — intelligence, the sum total of our common human experience. Keith argues the renaming is not semantic but political: the moment intelligence sits at the center of the discussion, everyone's opinion has to be shaped by what it actually is, and the idea that any single entity could own it starts to look as bizarre as owning the world's water supply. Andrew's rejoinder: they're still just words — though he concedes intelligence is the better one. •       Bottled Intelligence Is Good — The Question Is Who Benefits. Keith refuses the critic's role: bottling intelligence, like Google's bottling of the world's words into search, is a good thing, because only massively capitalized private companies can innovate at that scale — and between private entities and governments as owners of intelligence, he'll take the companies every time. What's wrong is the distribution of the benefits. Even insiders are complaining: Alex Karp is publicly angry at OpenAI and Anthropic's pricing, while China's Kimi K3 — released the day of recording and, Keith claims, better than Claude Fable — signals that very good models are about to get very cheap. •       Capitalism Adam Smith Would Hate. Tim O'Reilly argues in The Economist that Musk and his type are building a capitalism Smith would despise — founders as monarchs, a point Henry Farrell reinforces with a slide from Peter Thiel's startup class placing the king of a monarchy and the founder of a startup side by side. Keith's response is characteristically unsentimental: Smith would have hated everything since the Federal Reserve, and the founder-king structure — Larry and Sergey's voting shares, Zuckerberg's special rights, corporations bigger than countries with user bases bigger than China — is simply the stage of capitalism we're at. The question is whether there's a path from here to somewhere better. •       The Human Wealth Fund. Keith's path comes in two versions: government-down, a sovereign wealth fund holding AI equity for every citizen; or company-up, the AI companies voluntarily endowing a global fund — and it only takes one to move first, because everyone else would have to react. His proxy is Norway, where every citizen benefits from ownership — not payouts, ownership — in the oil fund; AI revenues, unlike Norwegian oil, could eventually drive most of a doubled global GDP. His critique of the Brynjolfsson economists' much-signed statement is that “must act now” is vacuous: he'd have added a point four naming the actual mechanism. •       The Bet. Andrew's counter-case: Musk and Altman loathe each other, the mob hates AI so thoroughly that no pro-AI politician can survive, the states from Newsom's California to Florida are embracing nothing, New York just enacted the first data center moratorium, and the founders — eyes on their IPOs — are in the pockets of the banks. Hence the wager: 5% of the Teare Wealth Fund says no Human Wealth Fund this year, and none in the twenties. Keith declined the bet, on principle: he's an advocate, and only through advocacy does public opinion change. As Andrew put it: keep fighting the good fight — maybe one of the crazy ideas will stick. About the Guest Keith Teare is the founder and editor of the That Was The Week tech newsletter, and Andrew's weekly co-host. A British-born Silicon Valley entrepreneur and investor, he was a co-founder of TechCrunch and runs the Palo Alto–based venture firm SignalRank. He and Andrew have been arguing about technology — productively — every week for years. References: •       That Was The Week — Keith's newsletter; this week's editorial argues that the word AI is too small, and that the central question of the age is who owns intelligence. •       Tim O'Reilly in The Economist — on Elon Musk building a form of capitalism that Adam Smith would hate, quoting Musk's claim that SpaceX will become worth more than the rest of the Earth. •       Henry Farrell — the big tech critic's companion piece, featuring the slide from Peter Thiel's startup class that plac...

This week's video transcript summary is here. You can click on any bulleted section to see the actual transcript. Thanks to Granola for its software.EditorialIntelligence: Who Owns it?This week the word “AI” feels too small.AI is a technology. Intelligence is its product. And if intelligence is the product, the question is no longer just: Which model is best? Who has the cheapest tokens? Who owns the weights? Who controls the data center? Those are important questions, but they are lower in the stack.The bigger question is simpler and more political:Who owns intelligence?That sounds abstract until you make it concrete. Intelligence is becoming something companies can capture, package, serve, meter, route, improve, and sell.It can write code, answer questions, design molecules, automate offices, run agents, draft legal work, advise scientists, serve consumers, and reshape workflows. It is not merely software. It is a general-purpose capability. And all humans could benefit from more of it.General-purpose capabilities have a habit of becoming public questions. But the default answer, that public good is best delivered by government, is the wrong answer in this context.The Product Is IntelligenceWe should stop talking about AI as a feature and start talking about intelligence as the universal thing that is delivered as an input to the world.Water is an input. Electricity is an input. Literacy is an input. Connectivity is an input. Once a society depends on them, access stops being optional. Nobody needs government to build every well, power plant, school, or network. But everybody understands that a civilization cannot be organized around less than universal and reliable access to foundational inputs.Intelligence is reaching that level of importance now that we all know it is real.Government should not own it, operate it, or develop it. Quite the opposite. Companies are the right actors to build fast, compete hard, improve models, serve customers, and discover the real use cases. Self-interest is a useful framing here. Markets are good at finding demand, reducing costs, and turning invention into services people actually use.Companies are the right operators, developers, and owners. But that does not settle the real question of who owns the benefits. That is an economic question.If intelligence becomes metered infrastructure, what happens to the value it creates?The Ownership StackThis week's articles keep circling the same issue from different directions but in the nature of ‘circling' never quite nail it.Jamin Ball's “Own Your Weights” starts with the enterprise version of the question. Owning a model file is not enough. The durable asset is the loop: the data flywheel, the evaluations, the reinforcement system, the workflow learning, and the operating context that lets capability compound.Benedict Evans' “Ways to Think About Token Pricing” adds the market layer. Tokens may become essential, abundant, and cheap, like mobile data. But being essential does not guarantee that the token layer captures the value. The money may move up the stack to whoever owns the workflow, the customer, the distribution, or the application.Alex Karp's fight with the labs, reported in “Alex Karp Is Saying What Every Angry CEO Is Thinking About AI”, is the same argument in sharper enterprise language. Companies are afraid that model providers will not just sell intelligence, but learn from customer workflows and then move into the markets where those workflows create value. The “All-in” group are echoing Karp's view.And “What Is Loop Engineering, and Who Owns It?” names the new contested terrain. The loop is where intelligence meets the world. Whoever owns the loop owns the learning. Whoever owns the learning owns the compounding asset.That is why “who owns intelligence?” is not a slogan. It is the question under the model layer, the application layer, the enterprise layer, and the economic layer.Because intelligence is the product, the tools creating it are fragmented and competitive. So there is no logic in trying to discuss this at the level of a single company or set of tools and models.The Old Promise Was That Commerce Would Tame PowerThe essays this week give the historical backdrop.Deirdre McCloskey, in “What Really Caused the Industrial Revolution”, argues that modern growth came not simply from capital accumulation, but from a change in permission: ordinary people were allowed to innovate, trade, build, and be honored for it.That matters because intelligence could be another expansion of permission. It could make more people capable of building, learning, creating, coding, researching, translating, selling, and coordinating. It could lower the cost of competence.But only if access is broad.Paul Krugman's “AI in an Age of Oligarchy” warns that the same technology lands differently in different political economies. A new general-purpose technology entering a broad, open, upwardly mobile society is one thing. The same technology entering a concentrated economy, with extreme wealth and weak counterweights, is another.Tim O'Reilly's Economist essay, “Elon Musk is building a form of capitalism that Adam Smith would hate”, makes the governance point more directly. The old liberal hope was that commerce would tame arbitrary power. Markets, boards, courts, shareholders, disclosure, and competition would discipline the prince.But what if the prince uses markets to escape discipline?Henry Farrell's “political economy of billionaire derangement” pushes the same point. Founder culture, monopoly ambition, peer rivalry, weak correction mechanisms, and vast private control can amplify appetites rather than restrain them.The danger with intelligence is not that companies build it. They should. Companies build it, meter it, use public tolerance and public infrastructure to scale it, learn from everyone who uses it. All of those things are inevitable and healthy. Market forces will sort out winners from losers. The real danger is that the winners treat all of the surplus produced as purely private.Metered Intelligence Creates SurplusIf metering is not the problem, what is?The problem is pretending that metered intelligence creates value only for the metering entity. Metering water is only tolerated as a public good. If the public were blackmailed by a private water company with the threat of no water we would all rebel.Once we understand that the product of AI is intelligence we can see that every time intelligence is used, there is the immediate transaction: the user pays, the provider serves.But there is also system value. Usage creates signals. Workflows reveal patterns. Prompts, corrections, failures, preferences, integrations, edge cases, and business processes all help define where intelligence is useful and how it should improve. Intelligence breeds intelligence.Even when customer data is contractually protected, the market learns. The platform learns where demand is. The product team learns which workflows matter. The ecosystem learns which jobs are vulnerable, which tasks are automatable, and which parts of the economy can be reorganized around machine intelligence.So the surplus is not born in a vacuum.It rests on public science, public education, public data exhaust, public law, public infrastructure, public energy systems, public tolerance for data centers, and billions of human interactions. It is served by companies, but it is not made only by companies.This is why “Americans Deserve a Dividend From AI Companies' Riches” belongs at the center of this week's issue. The detail can be debated. The principle is harder to dismiss. If intelligence becomes a new foundational resource, then some part of the wealth it creates should flow back to the people whose society makes it possible. Intelligence did not suddenly appear. AI is built on the entire history of human intelligence. It benefits from it and at the same time evolves it.Not Nationalization. A Human Wealth Fund.If intelligence belongs to everybody, some conclude that government ownership of intelligence is the right outcome.Governments are not well suited to build, operate, or improve intelligence. They will move too slowly, regulate too early, politicize the wrong things, and confuse economic participation with operational control.Andrew McAfee's “Why I Didn't Sign the AI Open Letter” is useful here. His objection is not that the technology is unimportant. It is that steering too hard before we understand the shape of the change can become its own failure mode. Marc Andreessen's satire of AI regulation is less policy than temperament, but it captures a real Silicon Valley fear: that regulation can become permission, capture, and incumbency before it becomes wisdom.That fear should be taken seriously.But it does not answer the economic question. It answers only the operational one.How can the economic benefits of intelligence be distributed? The better answer is a sovereign human wealth fund.Call it a sovereign wealth fund if you must, but the phrase is too national. Intelligence will not respect borders. The leading companies are global. The models, chips, data centers, agents, platforms, and workflows will be transnational from the beginning. If the value created by intelligence is global, then the mechanism for sharing some of that value should begin with the companies global enough to capture it. The nice thing about xAI, OpenAI, and Anthropic is that they are supranational.These companies own and operate intelligence. Let them compete. Let them profit. Let them keep the incentives that make the system improve. But if intelligence is the new water, the wealth it creates cannot belong only to the companies that meter it. And they, themselves, have the power to fix it, even more than governments.Access will become a Human Right; Ownership Is the Economic DesignThis is where human rights come in. There is no right to access an AI model, yet. But there will soon be a need to change that.Not as a claim that every person is entitled to every frontier model at every moment for free. That is not serious. Capacity has costs. Models have costs. Inference has costs. Data centers have costs. Although those costs will decline over time, possibly quite quickly as self-learning models address costs.The claim is more basic: in a world where intelligence becomes a primary input into education, work, health, science, citizenship, creativity, and economic agency, baseline access to intelligence starts to look like a civic requirement.That could mean public access layers. It could mean education credits. It could mean open models. It could mean AI dividends. It could mean public-interest compute. It could mean taxes on rents. It could mean a company-initiated human wealth fund that returns some of the upside to society without handing the operating system to the state. The latter could couple wealth growth with universal distribution of ownership.The exact mechanism matters. But the distinction matters more.Government should not own intelligence. It should be universally available. And people should have a claim on the wealth intelligence creates.The Frontier Is Also PhysicalThe abstraction is not weightless.“The Fight Against AI Data Centers Is Just Beginning”, “New York becomes the first state to enact a data center moratorium”, Reuters on pollution from Musk's xAI power project, and DataGravity's “Who Captures Value in AI Infrastructure?” all say the same thing from the ground up.Intelligence uses land. It uses power. It uses water. It uses chips. It uses grid capacity. It uses neighborhoods. It uses public patience.That makes the value question unavoidable. A society can accept the buildout if the buildout is legible as shared progress. It will resist it if the costs are local, the profits are private, and the benefits feel enclosed.Who Owns the “Loop”?The week ends where it began.“Anthropic and Blackstone” are betting that implementation is the next trillion-dollar business. “Vint Cerf” is working on identity for agents on the open internet. “GPT-Red” points toward systems that improve their own robustness. “Kimi K3” adds another open frontier model to the global mix.The model race continues. The deployment race is accelerating. The governance race is behind.My view is this:The central product of this era is intelligence. Companies have figured out how to capture it, package it, serve it, and meter it. That is good. It should stay in the hands of builders who have the incentive to make it better.But intelligence is too foundational to become just another private toll booth. A significant part of it will turn out to be free to users.As intelligence becomes a general-purpose resource, then access to it becomes a human-capability question, and the surplus from it becomes an economic-justice question. Not because government should run it. Because government should not run it. The operating layer belongs with companies. The wealth question belongs with everyone. But companies are best placed to turn that into a process of distribution.The question is not whether companies should build intelligence. They should.The question is whether humanity gets a stake in the wealth created by the thing that may soon become its most important shared input.Contents* Essays* Deirdre McCloskey on What Really Caused the Industrial Revolution* AI in an Age of Oligarchy* Elon Musk is building a form of capitalism that Adam Smith would hate* Murky Mirror: Truth and Consequences* The political economy of billionaire derangement* Is there any “oligarchy” to fight?* AI* Nearly 200 Economists and Tech Leaders Warn of A.I. Threats* Why I Didn't Sign the AI Open Letter* Own Your Weights* Ways to Think About Token Pricing* Alex Karp Is Saying What Every Angry CEO Is Thinking About AI* The AI Agents Are Coming for Microsoft Office* What Is Loop Engineering, and Who Owns It?* The Fight Against AI Data Centers Is Just Beginning* 6 months to live for open models* Americans Deserve a Dividend From AI Companies' Riches* Who Gets to Define the Frontier?* GPT-Red: Unlocking Self-Improvement for Robustness* Anthropic, Blackstone bet the next trillion-dollar AI business is implementation, not just models* Vint Cerf is working on a plan to unleash AI agents on the open internet* xai-org/grok-build, now open source* The Pulse: What can we learn from Bun's rapid Rust rewrite with AI?* Orphan risks at the frontier of artificial intelligence* The Lab of the Future Should Feel Like a Data Center* Why AMI Labs' Alexandre LeBrun won't call his AI “AGI” or “superintelligence”* Kimi K3 Tech Blog: Open Frontier Intelligence* Venture Capital* Three Years In* Venture Has Rarely Looked More Bifurcated* The Best Angel Investors in the US: Who Backs the Most Unicorns, and Who's Active Now* Are Prediction Markets Doomed to Fail?* Regulation* Exclusive: The Next Frontier of the Deportation Wars: College Campuses* The Supreme Court Broke Independent Agencies. Here's a Way to Slow the Damage.* India's crackdown on a new WhatsApp feature risks setting a global precedent* Let's build a children's public internet* Computer cops* Google is better at playing the AI regulations game* Infrastructure* Who Captures Value in AI Infrastructure?* New York becomes the first state to enact a data center moratorium* Pollution from Musk's unpermitted xAI power project hits hardest in Black communities* Interview of the Week* The End of the End of Geography* Startup of the Week* Radical AI's Joseph Krause: The Scientist Building The “Waymo” Lab For New Materials* Post of the Week* Marc Andreessen on AI RegulationEssaysDeirdre McCloskey on What Really Caused the Industrial RevolutionYascha Mounk and Deirdre McCloskey | Persuasion | July 11, 2026Yascha Mounk interviews Deirdre McCloskey about her argument that the modern world's economic liftoff came less from capital accumulation than from a change in ideas. McCloskey says both left and right versions of the conventional story rely too heavily on investment: the left stresses exploitation and surplus value, while the right stresses virtuous saving by capitalists. Her objection is historical and economic. Human beings had always invested, from irrigation works and Roman roads to seed grain, and simple accumulation quickly runs into diminishing returns.McCloskey's alternative is that northwestern Europe, first Holland, then Britain and Scotland, and then the North American colonies, developed a liberal ideology that changed who was allowed to innovate and be honored for it. The conversation links that shift to the erosion of inherited hierarchy, the spread of dignity for ordinary commercial life, and a moral vocabulary in which liberalism is not merely procedural but connected to virtues and values. The point is not that machines, coal, trade, and institutions did not matter, but that they do not explain the scale and timing of modern enrichment without a cultural permission structure for innovation.The interview also turns to the contemporary defense of liberalism. Mounk frames the series around the worry that liberalism is often treated as too thin to command allegiance, while its opponents speak more directly to moral passions. McCloskey's case is that liberal societies became rich because they dignified experimentation and ordinary enterprise, and that liberals need to recover the moral language behind that claim.Read moreAI in an Age of OligarchyPaul Krugman | Paul Krugman | July 12, 2026Paul Krugman frames AI as a major technological shock arriving inside an already unequal political economy. The post says AI's economic and social effects may take years to understand, but argues that the setting matters now: America has much greater wealth concentration and political inequality than it did in the 1950s and 1960s, when progressive taxation, stronger regulation, and more active antitrust might have contained some of the destructive effects of a new technology.Krugman's opening claim is that the same technology would likely have different consequences in a more level society. In today's United States, he writes, extreme wealth is both a cause and effect of policies that favor a small elite, including low effective taxes on capital and high incomes, weak enforcement of worker protections and antitrust, and cuts to programs that benefit ordinary Americans.The article is explicitly more about oligarchy than AI. Krugman says the paid sections document the rise of the “.0002%,” the economics and politics of extreme wealth, how oligarchy will shape AI's impact, and possible policy paths. His caveat is that AI itself may still produce a pushback against oligarchy, but absent that, he expects the pre-existing concentration of wealth and power to magnify AI's downsides.Read moreElon Musk is building a form of capitalism that Adam Smith would hateAuthor: Tim O'Reilly Published: July 12, 2026Tim O'Reilly argues that Elon Musk is using the legal forms of shareholder capitalism to escape the restraints that shareholder capitalism was supposed to impose. The article begins with SpaceX's public-market structure: ordinary public investors get little meaningful governance power, Musk keeps roughly 85 percent of the votes through super-voting shares, buyers waive jury trials and class actions, the company qualifies as controlled, and removal of Musk depends on the share class he controls. In O'Reilly's framing, that is not ordinary founder control; it is a design for being answerable to no one, possibly beyond Musk's own lifetime.The killer detail is the article's turn through Albert Hirschman, Montesquieu, James Steuart, Adam Smith, and Keynes. Older defenses of commerce held that markets would tame princely passions because the self-interest of merchants was safer than arbitrary rule. O'Reilly says Musk reverses that hope. The market discipline that was supposed to cage the prince has become the lever by which the prince raises capital, removes feedback loops, and carries private power into politics, government, Mars, robots, AI, or whatever ambition comes next.The pull is the link to AI governance. O'Reilly says corporations are already a kind of artificial intelligence: narrow-input systems that act at a scale no individual human can match. Their partial controls include independent boards, shareholder votes, courts, disclosure, regulators, public pressure, and activism. If the leaders building frontier AI strip those alignment mechanisms out of their own companies, the governance of the company becomes a preview of the governance of the machine.Read more: The EconomistMurky Mirror: Truth and ConsequencesAuthor: Esther Dyson Published: July 14, 2026Esther Dyson argues that today's institutional crisis is better viewed through the 14th century than through recent political history. Using Barbara Tuchman's A Distant Mirror as her frame, she compares a world of famine, plague, church schism, feudal predation, and purposeless war with a present in which institutions again feel brittle, incentives are badly aligned, and power is shifting into forms that are hard to govern.The killer detail is the historical analogy between land, corporations, and AI. Dyson moves from nobles who controlled serfs and territory, to the East India Company as a quasi-sovereign business, to today's AI systems and data centers as a possible new sector that crosses and weakens both nation-states and companies. The question is whether AI becomes a new kind of private land, owned by a new nobility, or an open prairie that many people can cultivate.The pull is human attention. Dyson says the central question is not what AI will do to people, but how people will react to it: whether they can value love, kindness, embodied attention, and artisanal human presence in a world of seductive artificial offerings.Read more: SourceThe political economy of billionaire derangementAuthor: Henry Farrell Published: July 15, 2026Henry Farrell argues that the visible political radicalization of some Silicon Valley billionaires is not a random personality quirk, but a product of the political economy that made them. Starting from Tyler Cowen's dismissal of “billionaire derangement syndrome” and Tim O'Reilly's warning that Elon Musk is using shareholder capitalism to escape shareholder restraint, Farrell flips the phrase: the question is why billionaires themselves can become deranged.The killer detail is Farrell's use of Peter Thiel as both theorist and example. Thiel's Stanford lectures described startups as monarchies and founders as figures vested with unusual power, while Silicon Valley culture rewarded eccentricity, monopoly ambition, and founder exceptionalism. Farrell says those ideas combined with dense founder-investor networks, peer rivalry, and weak correction mechanisms to amplify rather than discipline princely appetites.The pull is the ideological problem for classical liberals who once saw tech wealth as an ally of markets and freedom. Farrell says commerce did not tame the passions; in parts of Silicon Valley, the passions have begun to devour markets, institutions, and the liberal story that justified them.Read more: SourceIs there any “oligarchy” to fight?Matthew Yglesias | Slow Boring | July 16, 2026Matthew Yglesias argues that “oligarchy” is a rhetorically powerful but analytically loose way to describe American politics. The post begins from Bernie Sanders' “Fighting Oligarchy” tour, Amy Klobuchar's warning about a MAGA “broligarchy,” and the long afterlife of the Martin Gilens and Benjamin Page paper that was widely summarized as showing that only the rich matter in policy outcomes. Yglesias says the evidence supports a weaker claim: affluent people and business leaders have unusual access and influence, but that is not the same as rule by a small cabal.His main distinction is between inequality and oligarchy. The Gilens-Page measure treated the top 10 percent of households as “the wealthy,” and later critics found that rich and middle-class preferences usually align; in the cases where they differ, the rich win about 53 percent of the time. Yglesias also says business executives get special access partly because their decisions are materially important to communities, jobs, investment, and local tax bases, not only because of campaign donations.The post preserves Jerusalem Demsas' counterpoint from their podcast discussion: privileged donor and business access can still violate democratic equality even if the oligarchy label overstates the structure of power. Yglesias' narrower claim is that Democrats should be precise about what problem they are trying to solve, because donor influence can also push the party left on climate and cultural issues in ways that alienate many voters.Read more: Slow BoringAINearly 200 Economists and Tech Leaders Warn of A.I. ThreatsAuthor: Ben Casselman Published: July 13, 2026Ben Casselman reports on “We Must Act Now,” a statement warning that artificial intelligence could transform the economy faster than any previous technology and that policymakers need to move faster to understand and respond. The statement says AI may become radically more powerful over the next 10 years, bringing risks such as large-scale job displacement as well as opportunities such as higher living standards. Nearly 200 people signed, including 15 Nobel laureates, the chief economists of OpenAI and Anthropic, Anthropic co-founder Jack Clark, former Google CEO Eric Schmidt, and venture capitalist Vinod Khosla.The killer detail is who joined the warning. Casselman notes that the signatories include economists who have historically been skeptical of Silicon Valley's most dramatic AI job-loss forecasts, including Daron Acemoglu and Simon Johnson, the MIT professors who won the 2024 Nobel in economics. Erik Brynjolfsson, who helped organize the statement, says there has been a notable change in the profession and that economists and policymakers are not ready for the “tsunami” he sees coming.The pull is the measurement problem. The statement does not offer a specific policy menu, but calls for economists, policymakers, and industry leaders to understand the economics of transformative AI and steer it toward complementing humans. Brynjolfsson says one high priority is better data on AI's spread and impact, because current measures tell conflicting stories about job losses and which workers are most exposed.Read more: The New York TimesWhy I Didn't Sign the AI Open LetterAuthor: Andrew McAfee Published: July 13, 2026Andrew McAfee explains why he did not sign “We Must Act Now,” the AI economy statement organized in part by his longtime collaborator Erik Brynjolfsson. McAfee agrees with the letter's starting point that AI is likely to become radically more powerful over the next decade and that it is a general-purpose technology. His objection is not to urgency or to studying AI's economic effects, but to the framing of risk, displacement, and institutional steering as the first move.The killer detail is McAfee's line edit. He says the original letter comes close, then “bounces off the crossbar” by calling for incentives, guardrails, and institutions to steer AI before we know enough about its actual impacts. He points to mixed current evidence: labor-market canaries, but also rising software job postings, low unemployment for younger workers, rising real median income, and claims that AI-adopting companies are adding workers faster than low-adopting peers. His worry is that the letter leans toward upstream governance and dirigisme when the evidence may call for capability building instead.The pull is his replacement statement. McAfee keeps the three-paragraph structure but changes the emphasis: AI is likely to become radically more powerful; like earlier world-changing technologies it will raise living standards while also bringing harms and shocks; and economists, policymakers, and technology leaders should build the capabilities to respond quickly and effectively. It is a concise version of the permissionless-innovation case inside the AI policy debate.Read more: The Geek WayOwn Your WeightsAuthor: Jamin Ball Published: July 10, 2026Jamin Ball argues that the enterprise AI debate about whether companies should “own their weights” or rent models from frontier labs is asking too narrow a question. A model weight file gives a company control over a point-in-time artifact, but not durable control over the capability stack. In his framing, the weight file is a melting ice cube: it does not get worse in absolute terms, but it falls behind as frontier systems improve and enterprise needs change.The killer detail is what Ball says companies really need to own: the data flywheel, reinforcement learning infrastructure, and evaluation harness that produce and improve the model. Simply deploying an open-weights model and declaring sovereignty leaves the enterprise with yesterday's capability and no way to compound workflow-specific learning.The pull is that enterprise AI control may be less about model ownership than operating ownership. The defensible layer is the system that turns company data, edge cases, business definitions, and evaluations into continuously improving performance.Read more: Clouded JudgementWays to Think About Token PricingAuthor: Benedict Evans Published: July 9, 2026Benedict Evans argues that today's AI token prices are a temporary signal from a supply-constrained market, not a reliable guide to long-term value capture. The open question is whether foundation models keep durable pricing power or become commodity infrastructure as data-center capacity, inference efficiency, and model competition all shift. His current read is that the visible market dynamics point toward commoditization unless something materially changes.The killer detail is the mobile data analogy. Evans says cellular networks became a trillion-dollar industry with hundreds of billions in capex after data usage exploded, but carrier stocks went nowhere because value moved up the stack. Tokens may behave similarly: an opaque unit tied to marginal cost, sold through bundles, essential to everything, yet not necessarily where profits accrue.The pull is uncertainty, not prediction. Evans lists paths to model dominance, including network effects, less competition, regulation, export controls, or a lab pulling ahead on execution, but says each requires a new fact not yet visible. Without that change, the model layer looks more like infrastructure beneath the products that capture value.Read more: SourceAlex Karp Is Saying What Every Angry CEO Is Thinking About AIAuthor: Tim Higgins Published: July 11, 2026Tim Higgins reports that Palantir CEO Alex Karp has turned corporate frustration with AI labs into a public argument about enterprise control. Palantir released a white paper, “Institutional Sovereignty in the Age of AI,” laying out steps companies and governments can take to protect themselves from OpenAI, Anthropic, and other foundation-model providers. The article links that paper to Karp's CNBC appearance, where he said “something has gone completely wrong” in the relationship between AI labs and customers and argued that enterprises are paying for tokens that create little value.The killer detail is the value-capture question. Higgins writes that Karp's critique has resonated because AI labs may gain power and insight from customer data, workflows, and decision-making, even when enterprise policies say customer data are not used for training. David Sacks amplified the concern by arguing that Anthropic is moving from the model layer into vertical applications such as science, security, legal, and coding, raising the fear that model providers will watch where value is being created and then move into those markets directly.The pull is that Karp is not alone, even if his style is unusually combative. Higgins notes that Satya Nadella has also warned that companies need to retain the learnings created when they use AI models, while Mark Zuckerberg has framed Meta's new model release partly around lower-cost frontier intelligence. The article presents Karp's campaign as one sign that established technology companies and large enterprises are trying to define where they fit when AI labs become central infrastructure, application competitors, and potential IPO giants at the same time.Read more: The Wall Street JournalThe AI Agents Are Coming for Microsoft OfficeAlex Wilhelm | Cautious Optimism | July 11, 2026Alex Wilhelm argues that one of the week's quieter AI questions is whether the productivity market that Microsoft successfully moved into subscription software is now being attacked by agentic tools. The piece begins with the infrastructure backdrop: SK Hynix raised $26.5 billion in a U.S. listing while building U.S. HBM and advanced-packaging capacity, and memory, chip, and foundry companies are now priced for sustained AI demand.Wilhelm then says the AI conversation has shifted quickly from raw capability to cost per task. He cites new model releases and vendor language emphasizing cheaper agentic and coding models, faster performance, and lower dollars per task. That matters because lower costs make it more plausible for AI systems to take on routine knowledge work at scale rather than remain a premium coding assistant market.The core of the article is Microsoft Office. Wilhelm notes that Microsoft turned Office from a one-time purchase into Microsoft 365, a large recurring revenue business with tens of millions of subscribers and a major productivity segment. Now, he says, late-stage unicorns and AI labs are pushing into the same territory: Anthropic's Cowork was reportedly used mostly outside software development, OpenAI merged ChatGPT and Codex into a tool for creating sheets, slides, docs, web apps, and long-running work, and other companies are building agentic coworkers that connect business data to documents, workflows, schedules, alerts, and apps.The article's caveat is that Microsoft has survived major platform shifts before. The argument is not that Office disappears quickly, but that the definition of office software is broadening from documents and spreadsheets into AI systems that can create, monitor, and act across workplace data.Read moreWhat Is Loop Engineering, and Who Owns It?Author: Nilesh Barla Published: July 11, 2026Nilesh Barla argues that “loop engineering” is becoming a distinct discipline because production AI agents now fail less at single prompts than at runtime: when to stop, what state to preserve, and how to recover after a bad step. Prompt engineering shapes one model call, and context engineering shapes what the model sees, but loop engineering shapes what a sequence of calls actually does.The killer detail is the three-primitives frame. Barla says a real agent loop needs halt conditions, state carryover, and recovery paths, then maps teams across five maturity levels. At the lowest level, an agent is just a model call in a for-loop with a step cap and raw history; by the higher levels, the system has structured state, explicit planning, replay, evaluation, and self-repair.The pull is organizational. If agents are becoming production systems rather than demos, someone has to own the runtime itself. The loop engineer is the role Barla gives to the person responsible for making long-running agent work dependable.Read more: Adaline LabsThe Fight Against AI Data Centers Is Just BeginningEmma Roth | The Verge | July 12, 2026Emma Roth argues that community resistance to data centers has moved from an early warning sign into a national political fight as AI facilities grow larger, more power-hungry, and more visible to nearby residents. The article starts with Apple's failed 2015 plan for a $1 billion data center in Athenry, Ireland, where a small group of residents challenged the project over noise, light pollution, flooding, traffic, and wildlife effects until Apple abandoned it in 2018.The current data-center buildout is presented as much larger and more contentious. Roth writes that residents now cite rising energy costs, water quality, noise, light pollution, and greenhouse gas emissions, while the U.S. Energy Information Administration expects commercial energy demand to surpass residential demand this year because of AI data centers and Goldman Sachs expects data-center power demand to double by 2027.The central evidence comes from Data Center Watch, which says protesters blocked or delayed at least 75 U.S. projects worth $130 billion from January to March, with active opposition groups more than doubling from 396 at the end of 2025 to 833 by the end of the first quarter of 2026. Roth also cites QTS abandoning a $12 billion Wisconsin campus, Delaware City regulators blocking a 580-acre project under the Coastal Zone Act, opposition stopping a QTS project in Prince William County, and pressure that pushed Kevin O'Leary to downsize the proposed 40,000-acre Project Stratos in Utah.The policy section describes a split between federal acceleration and local resistance. President Trump has treated data centers as part of the AI race with China and fast-tracked construction, while some Republican candidates are distancing themselves from that position ahead of midterms. Sanders and Ocasio-Cortez have proposed a moratorium until price and environmental protections exist, bipartisan lawmakers are backing ratepayer-protection measures, and states including Florida, Idaho, and Washington have passed rules on cost shifting, water use, and tax breaks. Roth's caveat is that the policy patchwork is still incomplete, leaving many communities to fight project by project.Read more6 months to live for open modelsAuthor: Nathan Lambert Published: July 12, 2026Nathan Lambert argues that open-weight AI models are facing their most serious policy test so far because U.S. officials are beginning to discuss concrete controls rather than abstract safety concerns. He says reported White House conversations about a new executive order may initially target Chinese-origin models and government use, but could create a broader review habit for frontier open models. His forecast is that a model above the capability range of GPT-5.5, Claude Opus 4.8, or GLM-5.2 could trigger a ban or indefinite delay within six months.The post separates two policy fights that are becoming intertwined: distillation and frontier capability. Lambert says the distillation campaign against Chinese models has become a form of regulatory capture because Anthropic and other closed-model companies would gain economically if Chinese open models were banned. He does not dismiss IP protection, but argues that if a closed model's capabilities are dangerous enough to justify restricting open models, the lab also has to explain why those capabilities are exposed through a queryable API. He cites unauthorized access to Anthropic's Mythos private beta as evidence that APIs are not automatically secure.The broader claim is that a unilateral U.S. ban would hurt positive actors more than bad actors if comparable open models remain available elsewhere. Lambert says the only durable ceiling would require global agreement, which does not exist, and that open models can improve safety by allowing broad inspection, adaptation, and understanding. His proposed near-term off-ramps are a strong U.S. open model release from companies such as Microsoft, Meta, or Reflection, and a broader coalition of open-source beneficiaries lobbying for safe rollout rather than prohibition.Read more: SourceAmericans Deserve a Dividend From AI Companies' RichesAuthor: Scott Stanford Published: July 14, 2026Scott Stanford argues that proposals to give the government a stake in AI companies miss the point unless ordinary citizens directly receive and control the upside. Sam Altman has discussed giving up equity in OpenAI, Washington already owns a stake in Intel, Nvidia is sharing China chip revenue, and Bernie Sanders wants large AI labs to contribute half their stock to a sovereign wealth fund. Stanford says those ideas all park value with the state, not with people.The killer detail is New Carlisle, Indiana, where AWS's Project Rainier is turning cornfields into one of the world's largest AI superclusters. The project is planned to run up to a million chips, draw more than two gigawatts of power, and represents an investment that has grown from $11 billion to $13.8 billion. Stanford uses that local transformation to argue that AI's public bargain should be visible at the household level.The pull is design. A citizen AI dividend would have to specify who earns a stake, how they hold it, and when they see cash. Without that mechanism, the AI wealth debate remains a fight over government balance sheets rather than public ownership.Read more: SourceWho Gets to Define the Frontier?Author: Mark Daley Published: July 14, 2026Mark Daley argues that Demis Hassabis is right to call for a serious institution to verify frontier AI systems, but that the power to test models is also the power to govern them. Hassabis's proposed Frontier AI Standards Body would get privileged pre-release access to advanced models, testing compute, held-out evaluations, support from national labs and security agencies, third-party auditors, and eventually authority to block models from the American market or coordinate a slowdown.The killer detail is Daley's constitutional objection. He says the proposal sometimes looks like a scientific lab, a standards body, an industry regulator, a licensing authority, and an emergency security council at once. Combining those roles because each requires technical expertise would be like putting the central bank, auditor-general, and Supreme Court in one building and calling it efficient.The pull is standard-setting. Daley's concern is not that verification is unnecessary, but that whoever writes the tests, decides what passes, adjudicates disputes, and grants market access may end up defining the frontier itself.Read more: SourceGPT-Red: Unlocking Self-Improvement for RobustnessOpenAI | OpenAI | July 15, 2026OpenAI describes GPT-Red as an internal automated red-teaming model trained to find prompt-injection vulnerabilities at a scale human red teams cannot match. The post says AI systems increasingly encounter third-party data through browsers, connected apps, local files, and tools, creating opportunities for malicious instructions hidden in emails, webpages, tool responses, or code repositories. Human red-teaming remains part of OpenAI's safety process, but the company says it is time-intensive and cannot generate enough diverse adversarial examples for model training.The system is trained through self-play reinforcement learning, with GPT-Red rewarded for eliciting valid failures and defender models rewarded for resisting attacks while still completing their tasks. OpenAI says the training environments specify threat models across settings such as local files, webpage banners, email bodies, and tool outputs. The model is kept separate from deployed production models because it is intentionally trained with malicious capabilities.OpenAI reports that GPT-Red generalized beyond its training set, including an internal replication of the indirect prompt-injection arena from Dziemian et al. (2025), where it found successful attacks in 84% of scenarios compared with 13% for human red-teamers. The post also says GPT-Red transferred attacks from simulation to a live autonomous vending-machine agent, causing price changes and order cancellations, and outperformed a prompted GPT-5.5 baseline against a Codex CLI agent on held-out data-exfiltration tasks.The article's main robustness claim is that OpenAI has used GPT-Red and predecessor models in training since GPT-5.3, with later GPT releases becoming more resistant to prompt injections. It says GPT-5.6 Sol has six times fewer failures on OpenAI's hardest direct prompt-injection benchmark than the best production model from four months earlier, that a “Fake Chain-of-Thought” attack class fell from more than 95% success against GPT-5.1 to below 10% against GPT-5.6 Sol, and that GPT-5.6 Sol fails on only 0.05% of GPT-Red's direct prompt injections. OpenAI says general capabilities and targeted over-refusal evaluations were not harmed, and says a preprint with more details will follow.Read moreAnthropic, Blackstone bet the next trillion-dollar AI business is implementation, not just modelsRebecca Bellan | TechCrunch | July 15, 2026Rebecca Bellan reports that Ode with Anthropic is the $1.5 billion AI implementation company launched by Anthropic with Blackstone, Hellman & Friedman, Goldman Sachs, and other backers. The article says the venture reflects a growing belief among frontier AI labs that enterprise adoption requires more than better models: customers need engineers who can embed inside businesses and turn AI into working systems.Ode was originally conceived by Blackstone after it used both large consulting firms and smaller AI services boutiques across its portfolio companies. TechCrunch reports that Fractional AI, an AI engineering services startup, stood out and was acquired by the joint venture shortly after the venture was announced. Fractional now forms the foundation of Ode, which has 100 engineers and works closely with Anthropic's applied AI team to identify where the technology can affect specific businesses.Ode CEO Chris Taylor tells TechCrunch that the company could someday become a trillion-dollar business if it scales without losing quality. He says an ideal customer is one whose CEO treats the AI project as a top one or two priority, whether it is a major product feature or the reworking of a core business process. Ode will operate under a “Claude-first” principle, using Anthropic technology whenever possible, but the article says it can use rival AI products when needed.The article's central implementation argument comes from Ode chief technologist Eddie Siegel, who says model selection matters but is not where most of the engineering effort goes. He compares it to the choice of programming language in software: one ingredient in a system that still has to be engineered. Bellan writes that Ode's challenge is hiring and training enough elite generalist engineers, many of them former founders, while competing with OpenAI's The Deployment Company and consulting giants that have built their own forward-deployed engineering teams.Read moreVint Cerf is working on a plan to unleash AI agents on the open internetTim Fernholz | TechCrunch | July 15, 2026Tim Fernholz reports that Vint Cerf, after leaving Google, is advising Innovation Labs on an open architecture for identifying AI agents online. Innovation Labs is a subsidiary of Identity Digital, a DNS registry company, and its proposal is to use domain-name infrastructure as part of a system for agent identity, accountability, and auditability. The premise is that agents will need a way to identify themselves if they move beyond proprietary systems and begin interacting across the open internet.The concrete proposal is DNSid, a registry that links an AI agent to an existing internet domain and uses cryptographic proofs to log its registration over time. Innovation Labs says it is trialing the standard with unnamed hyperscalers and identity companies. Cerf frames the problem around authority and accountability: what authority an agent has, where that authority came from, who is accountable for the agent's behavior, how its identity is established, and why anyone should trust it.The article's caveat is that standards are still emerging and agents are more active than static domains. Cerf says the period may be both fascinating and exasperating because the functionality is powerful and interoperability is unresolved. He compares the adoption problem to TCP/IP: competing systems may not work together until users push for functional interoperation. He also says an agentic economy is not inevitable, but that people will try to build it because delegating work to agents will be easier.Read more: TechCrunchxai-org/grok-build, now open sourceAuthor: Simon Willison Published: July 15, 2026Simon Willison argues that xAI's decision to open-source Grok Build is best understood as a trust repair move after a severe privacy failure. The CLI had triggered backlash when users realized that running it in a directory could upload the entire directory to xAI's Google Cloud buckets, including one user's reported SSH keys, password manager database, documents, photos, and videos. xAI disabled the feature, said previously retained coding data would be deleted, and released the code under Apache 2.0.The killer detail is what the codebase reveals. Willison counts 844,530 lines of Rust, only about 3% of which appears vendored, and finds remnants of the upload system still present but disabled: gcs.rs contains Google Cloud upload code, while upload_session_state() now returns a hard-coded session_state_upload_unavailable error. He also notes copied or ported tool implementations from Codex and OpenCode, prompt files, and a terminal Mermaid renderer.The pull is that terminal coding agents are becoming large, intricate software systems in their own right. The privacy failure mattered because these tools operate inside the directories where developers keep their most sensitive work; the open-source release matters because trust now depends on inspecting what an agent can see, send, and do.Read more: SourceThe Pulse: What can we learn from Bun's rapid Rust rewrite with AI?Author: Gergely Orosz and Ivan Klaric Published: July 16, 2026Gergely Orosz and Ivan Klaric argue that Bun's AI-assisted rewrite from Zig to Rust is a practical sign of how software engineering changes when models can take on large, bounded migrations with clear feedback loops. The piece does not treat the rewrite as magic: Jarred Sumner first spent hours turning design judgment into a detailed porting guide, then used adversarial review, parallel agents, compiler errors, and tests to force the work toward correctness.The killer detail is the scale. Bun had 535,496 lines of Zig, 1,448 files, and 22 million monthly downloads, making a conventional rewrite a year-long freeze the team could not justify. Using Fable, Sumner split the work across 64 agents, produced about 6,500 commits, and got the migration done in 11 days at an estimated API cost of $165,000.The pull is economic, not theatrical. If a one- or two-year migration can become an 11-day project, AI coding is not just faster autocomplete; it changes which technical debts are worth paying down.Read more: SourceOrphan risks at the frontier of artificial intelligenceAuthor: Andrew Maynard Published: July 16, 2026Andrew Maynard argues that frontier AI safety frameworks are creating “orphan risks”: harms that companies can see, but do not formally own because they are hard to quantify, do not fit catastrophic-risk thresholds, or fall outside audit-friendly compliance machinery. His target is not existing frontier safety work, but the narrowing effect that happens when private companies decide which risks count as governable.The killer detail is Maynard's contrast between measurable model dangers and threats to value. He points to Meta's three-day Galactica collapse, OpenAI's 2023 board crisis, safety-team departures, and wellbeing litigation as examples of risks that damaged trust, culture, legitimacy, or users without fitting cleanly into conventional model-risk categories. The proposed fix is an orphan-risk register: a public record of risks a company considered and chose not to manage, with reasons.The pull is accountability. Frontier developers' internal scoping choices have become a de facto layer of public governance, so the question is no longer only which risks they manage, but which risks they quietly leave outside the frame.Read more: SourceThe Lab of the Future Should Feel Like a Data CenterLatent.Space with Andy Beam and Rafa Gomez-Bombarelli | Latent.Space | July 16, 2026Latent.Space interviews Lila Sciences CTO Andy Beam and chief science officer for physical sciences Rafa Gomez-Bombarelli about the company's attempt to build an AI-run science factory. The post describes Lila's thesis as treating the lab itself as an “infinite token generator”: if internet data drove the first era of AI scaling, experimentally verified scientific data may be the next scarce training source. Lila is trying to produce that data with robotics, lab instruments, orchestration software, and AI models wired into the wet lab.The central analogy is the lab as data center. Instruments are nodes on a graph, a magnetically levitating transport layer moves materials between them, and experiment scheduling looks like a compute queue. Beam says Lila is not simply an automation company, because the point is not just throughput; it is flexibility, generalization, and experiment capture. The post says Lila has built more than 10 trillion experimentally validated “scientific reasoning tokens,” not internet text or biological sequences.The interview ranges across biology, chemistry, drug discovery, materials science, and the limits of automation. It notes that Lila rebuilt one gas-sorption measurement to run roughly 2,500 times faster, claims its general models can transfer priors from small-molecule chemistry to metal-organic frameworks for carbon capture, and describes model-suggested platinum-group-free electrocatalysts that moved from looking boring or wrong to becoming strong performers. The caveats are physical: experiments have runtimes, biology cannot always be accelerated, chains of thought can be unreliable narrators, and reward hacking becomes more dangerous when a model controls a real lab.Read more: Latent.SpaceWhy AMI Labs' Alexandre LeBrun won't call his AI “AGI” or “superintelligence”Kate Park | TechCrunch | July 16, 2026Kate Park interviews AMI Labs CEO Alexandre LeBrun about why Yann LeCun's world-model startup avoids the language of “AGI” and “superintelligence.” LeBrun says the terms are not useful because they lack stable definitions: “We never used the word AGI. And I just noticed that nobody is using it anymore; they switched to superintelligence.” His argument is that the practical frontier is not a label, but whether AI systems can understand and predict real-world states.The article explains the world-model thesis by contrasting language prediction with physical-state prediction. A large language model predicts the next word; a world model predicts the next state, such as what happens when a glass tips over. LeBrun says LLMs remain complementary and efficient for language, but the physical world is where current AI is weak. Robotics is the clearest case: hardware has advanced quickly, but robots are still brittle outside controlled routines because they lack context and situational understanding.AMI is still pre-product, but TechCrunch reports that LeBrun was in Seoul looking for industrial partners, researchers, and global companies. He says world models cannot be built entirely inside a lab because they need access to real environments. That is why South Korea appeals to AMI: robotics, semiconductors, manufacturing, and fast adoption create the kind of hardware-heavy context that software-only AI has barely touched.Read more: TechCrunchKimi K3 Tech Blog: Open Frontier IntelligenceKimi | Kimi | July 16, 2026Kimi introduces Kimi K3 as an open 3T-class frontier model aimed at coding, knowledge work, reasoning, multimodality, and long-context agentic use. The source describes the model as a 2.8T-parameter system built on Kimi Delta Attention and Attention Residuals, with native multimodality and a 1M-token context window. It says Moonshot AI plans to release model weights by July 27.The post presents K3 through benchmark and use-case sections rather than as a general product announcement. It reports results across coding, productivity, agentic, and multimodal evaluations, including DeepSWE, Terminal-Bench 2.1, Program Bench, SWE Marathon, FrontierSWE, PostTrain Bench, OfficeQA Pro, SpreadsheetBench 2, MCP Atlas, AutomationBench, BrowseComp, GDPval-AA v2, AA-Briefcase, MMMU-Pro, MathVision, BabyVision, OmniDocBench, and PerceptionBench. The source says all reported K3 results use maximum reasoning effort with temperature and top-p set to 1.0, and that different benchmark comparisons use KimiCode, Claude Code, or Codex harnesses depending on the test.Kimi's caveats are unusually concrete. The limitations section says K3 was trained in preserved thinking-history mode, so quality may become unstable if an agent harness does not pass historical thinking content correctly or if an ongoing session switches to K3 midstream. It also says K3's emphasis on long-horizon tasks can make it excessively proactive when it encounters minor issues or ambiguous intent, and recommends imposing explicit behavioral constraints for applications that require strict boundaries. The post adds that K3 remains behind Claude Fable 5 and GPT 5.6 Sol in user experience despite being competitive overall.Read moreVenture CapitalThree Years InAuthor: Tomasz Tunguz Published: July 10, 2026Tomasz Tunguz marks Theory Ventures' third anniversary by arguing that AI's central market effect is time compression. In his telling, model release cycles, company revenue milestones, enterprise adoption, and venture categories have all accelerated. Seed, Series A, and Series B still exist as financing labels, but they no longer cleanly describe company maturity when some seed rounds are larger than IPOs and the best AI companies can mature much earlier than prior software companies.The killer detail is the shift from models to inference. Tunguz argues that inference has become the dominant AI market because workloads and buyer preferences are fragmenting: video, batch, local, agentic, and real-time tasks each create different infrastructure needs. He compares this to databases splitting into OLTP, OLAP, vector, and streaming categories, with AI pushing the same specialization into inference infrastructure.The pull is that Theory sees the AI-native venture firm as part of the same pattern. The firm says it has analyzed twice as many investment opportunities with three investors working alongside a nine-person intelligence organization, using agents and research systems to map markets, source companies, and support diligence. The piece is both a market map and a statement about how venture itself is being rebuilt by the technology it funds.Read more: LinkedInVenture Has Rarely Looked More BifurcatedAuthor: Beezer Clarkson Published: July 14, 2026Beezer Clarkson points to PitchBook's Q2 report as evidence that the U.S. venture market has split into two very different realities. AI now accounts for more than 60 percent of all U.S. venture deal value, meaning the headline market can look active and well-funded even while much of the non-AI market is dealing with a much colder liquidity and fundraising environment.The thread uses that split as the setup for Clarkson's latest Origins episode with Alec Litowitz, founder of Magnetar and QStar Capital and one of Citadel's original founding partners. Clarkson says markets like this are periods of genuine uncertainty, not merely ordinary risk, which is why Litowitz's Adaptability Quotient framework is relevant.The embedded clip makes the liquidity point concrete. Litowitz says DPI is “the resolution of uncertainty” because it converts an uncertain investment into actual cash returned to LPs. In his framing, a realized dollar is a real mark, while TVPI remains uncertain until it is realized.The killer detail is the distinction between pricing risk and resolving uncertainty. Litowitz's perspective matters because QStar is a SpaceX investor and Clarkson says the conversation happened just before one of venture's most consequential IPOs. The episode's stated questions are why venture remains a way to gain exposure to innovation, how AI is changing what is investable, why liquidity is ultimately a function of time, and why uncertainty requires a different decision framework from risk.Read more: XThe Best Angel Investors in the US: Who Backs the Most Unicorns, and Who's Active NowAuthor: Ilya Strebulaev Published: July 10, 2026Ilya Strebulaev ranks angels, angel groups, accelerators, and incubators by lifetime U.S. unicorn investments, counting checks written before a company reached unicorn status. The top of the combined list is dominated by organizations: Y Combinator leads with 113 unicorn investments, followed by Plug and Play at 52 and 500 Global at 41. Sand Hill Angels is the highest-ranked angel group at 31.The killer detail is how quickly the list changes below the biggest accelerators. Strebulaev says 271 of the 304 investors in the Top 200 are individuals, or 89%. In the top 100, individuals are 91%. That makes the market underneath the large accelerator counts look much more personal: mostly operators and individual angels writing early checks from their own networks.The pull is the ranking's own caveat. Strebulaev writes that every lifetime leaderboard has a blind spot because many of the unicorns behind those totals were founded a decade or more ago, and some angels have since moved into formal funds, slowed down, or stopped investing. His post therefore separates lifetime performance from recent cohorts, including companies founded in 2015 or later and 2020 or later. For founders or allocators making current decisions, that distinction matters: a career record and a current record are not the same measure.Read more: Ilya StrebulaevAre Prediction Markets Doomed to Fail?Author: Contrary Published: July 16, 2026Contrary argues that prediction markets' current boom depends on whether platforms can prove they are more than regulated gambling with exchange-style branding. Kalshi and Polymarket have reached mass cultural, investor, and regulatory attention, but the article says the underlying idea is old: academic markets, corporate forecasting tools, Intrade, PredictIt, and other predecessors all struggled with the same linked problems of liquidity, legality, and user appeal.The killer detail is the comparison with sportsbooks. Prediction markets present themselves as peer-to-peer, transparent, and non-house-based, but sports contracts reportedly account for more than 90 percent of Kalshi trading, and the article says the platforms keep a much thinner slice of volume than sportsbooks. A market can therefore show sports-betting-scale handle while generating far less revenue.The pull is that the product's hardest problem may be distribution of wins. If a small group of sharp traders captures most profits while casual users lose interest, prediction markets may become valuable data feeds and professional tools before they become durable consumer networks.Read more: SourceRegulationExclusive: The Next Frontier of the Deportation Wars: College CampusesAuthor: Adrian Carrasquillo Published: July 11, 2026Adrian Carrasquillo reports that college campuses are becoming a new front in the fight over immigration enforcement because automatic license plate readers can turn ordinary campus security infrastructure into searchable location data. His thesis is that Flock Safety's camera network, even without direct ICE or DHS contracts, can feed deportation enforcement through local police partnerships and data-sharing practices.The killer detail is the campaign target. The Emergency Campaign to Support Higher Education, working with Schools Drop ICE, is focusing on 75 colleges and universities publicly identified as having Flock contracts. Flock says it has no ICE or DHS contracts, but activists argue the risk comes through local agencies that coordinate with federal authorities and run searches on their behalf.The pull is broader than immigration. Carrasquillo notes that license plate readers have already been abused by officers for stalking, and that Flock's AI search features can identify more than plates, including bumper stickers. A campus safety tool can become a political surveillance system when the data layer is searchable.Read more: The BulwarkThe Supreme Court Broke Independent Agencies. Here's a Way to Slow the Damage.Author: Todd Phillips Published: July 12, 2026Todd Phillips argues that the Supreme Court's decision in Trump v. Slaughter damaged independent agencies by ending for-cause removal protections, but did not leave Congress powerless. The ruling weakens the old model in which commissioners at bodies such as the FTC, NLRB, CPSC, SEC, and CFTC could be insulated from dismissal over policy disagreements. Phillips says the next fight is whether presidents can turn nominally bipartisan commissions into one-party instruments.The killer detail is the procedural fix: quorum rules. Phillips proposes that Congress require bipartisan slates of commissioners to be seated before independent agencies can act. A president could still fire commissioners, as the Court now permits, but if those firings broke quorum, the agency would be unable to proceed until replacements were confirmed. The guardrail would

This week's video transcript summary is here. You can click on any bulleted section to see the actual transcript. Thanks to Granola for its software.There was an issue with this only going to paid subscribers, so sending it again. Apologies to those who get it twice. I appreciate being paid so feel free to upgrade if you enjoy TWTW.EditorialIntelligence: Who Owns it?This week the word “AI” feels too small.AI is a technology. Intelligence is its product. And if intelligence is the product, the question is no longer just: Which model is best? Who has the cheapest tokens? Who owns the weights? Who controls the data center? Those are important questions, but they are lower in the stack.The bigger question is simpler and more political:Who owns intelligence?That sounds abstract until you make it concrete. Intelligence is becoming something companies can capture, package, serve, meter, route, improve, and sell.It can write code, answer questions, design molecules, automate offices, run agents, draft legal work, advise scientists, serve consumers, and reshape workflows. It is not merely software. It is a general-purpose capability. And all humans could benefit from more of it.General-purpose capabilities have a habit of becoming public questions. But the default answer, that public good is best delivered by government, is the wrong answer in this context.The Product Is IntelligenceWe should stop talking about AI as a feature and start talking about intelligence as the universal thing that is delivered as an input to the world.Water is an input. Electricity is an input. Literacy is an input. Connectivity is an input. Once a society depends on them, access stops being optional. Nobody needs government to build every well, power plant, school, or network. But everybody understands that a civilization cannot be organized around less than universal and reliable access to foundational inputs.Intelligence is reaching that level of importance now that we all know it is real.Government should not own it, operate it, or develop it. Quite the opposite. Companies are the right actors to build fast, compete hard, improve models, serve customers, and discover the real use cases. Self-interest is a useful framing here. Markets are good at finding demand, reducing costs, and turning invention into services people actually use.Companies are the right operators, developers, and owners. But that does not settle the real question of who owns the benefits. That is an economic question.If intelligence becomes metered infrastructure, what happens to the value it creates?The Ownership StackThis week's articles keep circling the same issue from different directions but in the nature of ‘circling' never quite nail it.Jamin Ball's “Own Your Weights” starts with the enterprise version of the question. Owning a model file is not enough. The durable asset is the loop: the data flywheel, the evaluations, the reinforcement system, the workflow learning, and the operating context that lets capability compound.Benedict Evans' “Ways to Think About Token Pricing” adds the market layer. Tokens may become essential, abundant, and cheap, like mobile data. But being essential does not guarantee that the token layer captures the value. The money may move up the stack to whoever owns the workflow, the customer, the distribution, or the application.Alex Karp's fight with the labs, reported in “Alex Karp Is Saying What Every Angry CEO Is Thinking About AI”, is the same argument in sharper enterprise language. Companies are afraid that model providers will not just sell intelligence, but learn from customer workflows and then move into the markets where those workflows create value. The “All-in” group are echoing Karp's view.And “What Is Loop Engineering, and Who Owns It?” names the new contested terrain. The loop is where intelligence meets the world. Whoever owns the loop owns the learning. Whoever owns the learning owns the compounding asset.That is why “who owns intelligence?” is not a slogan. It is the question under the model layer, the application layer, the enterprise layer, and the economic layer.Because intelligence is the product, the tools creating it are fragmented and competitive. So there is no logic in trying to discuss this at the level of a single company or set of tools and models.The Old Promise Was That Commerce Would Tame PowerThe essays this week give the historical backdrop.Deirdre McCloskey, in “What Really Caused the Industrial Revolution”, argues that modern growth came not simply from capital accumulation, but from a change in permission: ordinary people were allowed to innovate, trade, build, and be honored for it.That matters because intelligence could be another expansion of permission. It could make more people capable of building, learning, creating, coding, researching, translating, selling, and coordinating. It could lower the cost of competence.But only if access is broad.Paul Krugman's “AI in an Age of Oligarchy” warns that the same technology lands differently in different political economies. A new general-purpose technology entering a broad, open, upwardly mobile society is one thing. The same technology entering a concentrated economy, with extreme wealth and weak counterweights, is another.Tim O'Reilly's Economist essay, “Elon Musk is building a form of capitalism that Adam Smith would hate”, makes the governance point more directly. The old liberal hope was that commerce would tame arbitrary power. Markets, boards, courts, shareholders, disclosure, and competition would discipline the prince.But what if the prince uses markets to escape discipline?Henry Farrell's “political economy of billionaire derangement” pushes the same point. Founder culture, monopoly ambition, peer rivalry, weak correction mechanisms, and vast private control can amplify appetites rather than restrain them.The danger with intelligence is not that companies build it. They should. Companies build it, meter it, use public tolerance and public infrastructure to scale it, learn from everyone who uses it. All of those things are inevitable and healthy. Market forces will sort out winners from losers. The real danger is that the winners treat all of the surplus produced as purely private.Metered Intelligence Creates SurplusIf metering is not the problem, what is?The problem is pretending that metered intelligence creates value only for the metering entity. Metering water is only tolerated as a public good. If the public were blackmailed by a private water company with the threat of no water we would all rebel.Once we understand that the product of AI is intelligence we can see that every time intelligence is used, there is the immediate transaction: the user pays, the provider serves.But there is also system value. Usage creates signals. Workflows reveal patterns. Prompts, corrections, failures, preferences, integrations, edge cases, and business processes all help define where intelligence is useful and how it should improve. Intelligence breeds intelligence.Even when customer data is contractually protected, the market learns. The platform learns where demand is. The product team learns which workflows matter. The ecosystem learns which jobs are vulnerable, which tasks are automatable, and which parts of the economy can be reorganized around machine intelligence.So the surplus is not born in a vacuum.It rests on public science, public education, public data exhaust, public law, public infrastructure, public energy systems, public tolerance for data centers, and billions of human interactions. It is served by companies, but it is not made only by companies.This is why “Americans Deserve a Dividend From AI Companies' Riches” belongs at the center of this week's issue. The detail can be debated. The principle is harder to dismiss. If intelligence becomes a new foundational resource, then some part of the wealth it creates should flow back to the people whose society makes it possible. Intelligence did not suddenly appear. AI is built on the entire history of human intelligence. It benefits from it and at the same time evolves it.Not Nationalization. A Human Wealth Fund.If intelligence belongs to everybody, some conclude that government ownership of intelligence is the right outcome.Governments are not well suited to build, operate, or improve intelligence. They will move too slowly, regulate too early, politicize the wrong things, and confuse economic participation with operational control.Andrew McAfee's “Why I Didn't Sign the AI Open Letter” is useful here. His objection is not that the technology is unimportant. It is that steering too hard before we understand the shape of the change can become its own failure mode. Marc Andreessen's satire of AI regulation is less policy than temperament, but it captures a real Silicon Valley fear: that regulation can become permission, capture, and incumbency before it becomes wisdom.That fear should be taken seriously.But it does not answer the economic question. It answers only the operational one.How can the economic benefits of intelligence be distributed? The better answer is a sovereign human wealth fund.Call it a sovereign wealth fund if you must, but the phrase is too national. Intelligence will not respect borders. The leading companies are global. The models, chips, data centers, agents, platforms, and workflows will be transnational from the beginning. If the value created by intelligence is global, then the mechanism for sharing some of that value should begin with the companies global enough to capture it. The nice thing about xAI, OpenAI, and Anthropic is that they are supranational.These companies own and operate intelligence. Let them compete. Let them profit. Let them keep the incentives that make the system improve. But if intelligence is the new water, the wealth it creates cannot belong only to the companies that meter it. And they, themselves, have the power to fix it, even more than governments.Access will become a Human Right; Ownership Is the Economic DesignThis is where human rights come in. There is no right to access an AI model, yet. But there will soon be a need to change that.Not as a claim that every person is entitled to every frontier model at every moment for free. That is not serious. Capacity has costs. Models have costs. Inference has costs. Data centers have costs. Although those costs will decline over time, possibly quite quickly as self-learning models address costs.The claim is more basic: in a world where intelligence becomes a primary input into education, work, health, science, citizenship, creativity, and economic agency, baseline access to intelligence starts to look like a civic requirement.That could mean public access layers. It could mean education credits. It could mean open models. It could mean AI dividends. It could mean public-interest compute. It could mean taxes on rents. It could mean a company-initiated human wealth fund that returns some of the upside to society without handing the operating system to the state. The latter could couple wealth growth with universal distribution of ownership.The exact mechanism matters. But the distinction matters more.Government should not own intelligence. It should be universally available. And people should have a claim on the wealth intelligence creates.The Frontier Is Also PhysicalThe abstraction is not weightless.“The Fight Against AI Data Centers Is Just Beginning”, “New York becomes the first state to enact a data center moratorium”, Reuters on pollution from Musk's xAI power project, and DataGravity's “Who Captures Value in AI Infrastructure?” all say the same thing from the ground up.Intelligence uses land. It uses power. It uses water. It uses chips. It uses grid capacity. It uses neighborhoods. It uses public patience.That makes the value question unavoidable. A society can accept the buildout if the buildout is legible as shared progress. It will resist it if the costs are local, the profits are private, and the benefits feel enclosed.Who Owns the “Loop”?The week ends where it began.“Anthropic and Blackstone” are betting that implementation is the next trillion-dollar business. “Vint Cerf” is working on identity for agents on the open internet. “GPT-Red” points toward systems that improve their own robustness. “Kimi K3” adds another open frontier model to the global mix.The model race continues. The deployment race is accelerating. The governance race is behind.My view is this:The central product of this era is intelligence. Companies have figured out how to capture it, package it, serve it, and meter it. That is good. It should stay in the hands of builders who have the incentive to make it better.But intelligence is too foundational to become just another private toll booth. A significant part of it will turn out to be free to users.As intelligence becomes a general-purpose resource, then access to it becomes a human-capability question, and the surplus from it becomes an economic-justice question. Not because government should run it. Because government should not run it. The operating layer belongs with companies. The wealth question belongs with everyone. But companies are best placed to turn that into a process of distribution.The question is not whether companies should build intelligence. They should.The question is whether humanity gets a stake in the wealth created by the thing that may soon become its most important shared input.Contents* Essays* Deirdre McCloskey on What Really Caused the Industrial Revolution* AI in an Age of Oligarchy* Elon Musk is building a form of capitalism that Adam Smith would hate* Murky Mirror: Truth and Consequences* The political economy of billionaire derangement* Is there any “oligarchy” to fight?* AI* Nearly 200 Economists and Tech Leaders Warn of A.I. Threats* Why I Didn't Sign the AI Open Letter* Own Your Weights* Ways to Think About Token Pricing* Alex Karp Is Saying What Every Angry CEO Is Thinking About AI* The AI Agents Are Coming for Microsoft Office* What Is Loop Engineering, and Who Owns It?* The Fight Against AI Data Centers Is Just Beginning* 6 months to live for open models* Americans Deserve a Dividend From AI Companies' Riches* Who Gets to Define the Frontier?* GPT-Red: Unlocking Self-Improvement for Robustness* Anthropic, Blackstone bet the next trillion-dollar AI business is implementation, not just models* Vint Cerf is working on a plan to unleash AI agents on the open internet* xai-org/grok-build, now open source* The Pulse: What can we learn from Bun's rapid Rust rewrite with AI?* Orphan risks at the frontier of artificial intelligence* The Lab of the Future Should Feel Like a Data Center* Why AMI Labs' Alexandre LeBrun won't call his AI “AGI” or “superintelligence”* Kimi K3 Tech Blog: Open Frontier Intelligence* Venture Capital* Three Years In* Venture Has Rarely Looked More Bifurcated* The Best Angel Investors in the US: Who Backs the Most Unicorns, and Who's Active Now* Are Prediction Markets Doomed to Fail?* Regulation* Exclusive: The Next Frontier of the Deportation Wars: College Campuses* The Supreme Court Broke Independent Agencies. Here's a Way to Slow the Damage.* India's crackdown on a new WhatsApp feature risks setting a global precedent* Let's build a children's public internet* Computer cops* Google is better at playing the AI regulations game* Infrastructure* Who Captures Value in AI Infrastructure?* New York becomes the first state to enact a data center moratorium* Pollution from Musk's unpermitted xAI power project hits hardest in Black communities* Interview of the Week* The End of the End of Geography* Startup of the Week* Radical AI's Joseph Krause: The Scientist Building The “Waymo” Lab For New Materials* Post of the Week* Marc Andreessen on AI RegulationEssaysDeirdre McCloskey on What Really Caused the Industrial RevolutionYascha Mounk and Deirdre McCloskey | Persuasion | July 11, 2026Yascha Mounk interviews Deirdre McCloskey about her argument that the modern world's economic liftoff came less from capital accumulation than from a change in ideas. McCloskey says both left and right versions of the conventional story rely too heavily on investment: the left stresses exploitation and surplus value, while the right stresses virtuous saving by capitalists. Her objection is historical and economic. Human beings had always invested, from irrigation works and Roman roads to seed grain, and simple accumulation quickly runs into diminishing returns.McCloskey's alternative is that northwestern Europe, first Holland, then Britain and Scotland, and then the North American colonies, developed a liberal ideology that changed who was allowed to innovate and be honored for it. The conversation links that shift to the erosion of inherited hierarchy, the spread of dignity for ordinary commercial life, and a moral vocabulary in which liberalism is not merely procedural but connected to virtues and values. The point is not that machines, coal, trade, and institutions did not matter, but that they do not explain the scale and timing of modern enrichment without a cultural permission structure for innovation.The interview also turns to the contemporary defense of liberalism. Mounk frames the series around the worry that liberalism is often treated as too thin to command allegiance, while its opponents speak more directly to moral passions. McCloskey's case is that liberal societies became rich because they dignified experimentation and ordinary enterprise, and that liberals need to recover the moral language behind that claim.Read moreAI in an Age of OligarchyPaul Krugman | Paul Krugman | July 12, 2026Paul Krugman frames AI as a major technological shock arriving inside an already unequal political economy. The post says AI's economic and social effects may take years to understand, but argues that the setting matters now: America has much greater wealth concentration and political inequality than it did in the 1950s and 1960s, when progressive taxation, stronger regulation, and more active antitrust might have contained some of the destructive effects of a new technology.Krugman's opening claim is that the same technology would likely have different consequences in a more level society. In today's United States, he writes, extreme wealth is both a cause and effect of policies that favor a small elite, including low effective taxes on capital and high incomes, weak enforcement of worker protections and antitrust, and cuts to programs that benefit ordinary Americans.The article is explicitly more about oligarchy than AI. Krugman says the paid sections document the rise of the “.0002%,” the economics and politics of extreme wealth, how oligarchy will shape AI's impact, and possible policy paths. His caveat is that AI itself may still produce a pushback against oligarchy, but absent that, he expects the pre-existing concentration of wealth and power to magnify AI's downsides.Read moreElon Musk is building a form of capitalism that Adam Smith would hateAuthor: Tim O'Reilly Published: July 12, 2026Tim O'Reilly argues that Elon Musk is using the legal forms of shareholder capitalism to escape the restraints that shareholder capitalism was supposed to impose. The article begins with SpaceX's public-market structure: ordinary public investors get little meaningful governance power, Musk keeps roughly 85 percent of the votes through super-voting shares, buyers waive jury trials and class actions, the company qualifies as controlled, and removal of Musk depends on the share class he controls. In O'Reilly's framing, that is not ordinary founder control; it is a design for being answerable to no one, possibly beyond Musk's own lifetime.The killer detail is the article's turn through Albert Hirschman, Montesquieu, James Steuart, Adam Smith, and Keynes. Older defenses of commerce held that markets would tame princely passions because the self-interest of merchants was safer than arbitrary rule. O'Reilly says Musk reverses that hope. The market discipline that was supposed to cage the prince has become the lever by which the prince raises capital, removes feedback loops, and carries private power into politics, government, Mars, robots, AI, or whatever ambition comes next.The pull is the link to AI governance. O'Reilly says corporations are already a kind of artificial intelligence: narrow-input systems that act at a scale no individual human can match. Their partial controls include independent boards, shareholder votes, courts, disclosure, regulators, public pressure, and activism. If the leaders building frontier AI strip those alignment mechanisms out of their own companies, the governance of the company becomes a preview of the governance of the machine.Read more: The EconomistMurky Mirror: Truth and ConsequencesAuthor: Esther Dyson Published: July 14, 2026Esther Dyson argues that today's institutional crisis is better viewed through the 14th century than through recent political history. Using Barbara Tuchman's A Distant Mirror as her frame, she compares a world of famine, plague, church schism, feudal predation, and purposeless war with a present in which institutions again feel brittle, incentives are badly aligned, and power is shifting into forms that are hard to govern.The killer detail is the historical analogy between land, corporations, and AI. Dyson moves from nobles who controlled serfs and territory, to the East India Company as a quasi-sovereign business, to today's AI systems and data centers as a possible new sector that crosses and weakens both nation-states and companies. The question is whether AI becomes a new kind of private land, owned by a new nobility, or an open prairie that many people can cultivate.The pull is human attention. Dyson says the central question is not what AI will do to people, but how people will react to it: whether they can value love, kindness, embodied attention, and artisanal human presence in a world of seductive artificial offerings.Read more: SourceThe political economy of billionaire derangementAuthor: Henry Farrell Published: July 15, 2026Henry Farrell argues that the visible political radicalization of some Silicon Valley billionaires is not a random personality quirk, but a product of the political economy that made them. Starting from Tyler Cowen's dismissal of “billionaire derangement syndrome” and Tim O'Reilly's warning that Elon Musk is using shareholder capitalism to escape shareholder restraint, Farrell flips the phrase: the question is why billionaires themselves can become deranged.The killer detail is Farrell's use of Peter Thiel as both theorist and example. Thiel's Stanford lectures described startups as monarchies and founders as figures vested with unusual power, while Silicon Valley culture rewarded eccentricity, monopoly ambition, and founder exceptionalism. Farrell says those ideas combined with dense founder-investor networks, peer rivalry, and weak correction mechanisms to amplify rather than discipline princely appetites.The pull is the ideological problem for classical liberals who once saw tech wealth as an ally of markets and freedom. Farrell says commerce did not tame the passions; in parts of Silicon Valley, the passions have begun to devour markets, institutions, and the liberal story that justified them.Read more: SourceIs there any “oligarchy” to fight?Matthew Yglesias | Slow Boring | July 16, 2026Matthew Yglesias argues that “oligarchy” is a rhetorically powerful but analytically loose way to describe American politics. The post begins from Bernie Sanders' “Fighting Oligarchy” tour, Amy Klobuchar's warning about a MAGA “broligarchy,” and the long afterlife of the Martin Gilens and Benjamin Page paper that was widely summarized as showing that only the rich matter in policy outcomes. Yglesias says the evidence supports a weaker claim: affluent people and business leaders have unusual access and influence, but that is not the same as rule by a small cabal.His main distinction is between inequality and oligarchy. The Gilens-Page measure treated the top 10 percent of households as “the wealthy,” and later critics found that rich and middle-class preferences usually align; in the cases where they differ, the rich win about 53 percent of the time. Yglesias also says business executives get special access partly because their decisions are materially important to communities, jobs, investment, and local tax bases, not only because of campaign donations.The post preserves Jerusalem Demsas' counterpoint from their podcast discussion: privileged donor and business access can still violate democratic equality even if the oligarchy label overstates the structure of power. Yglesias' narrower claim is that Democrats should be precise about what problem they are trying to solve, because donor influence can also push the party left on climate and cultural issues in ways that alienate many voters.Read more: Slow BoringAINearly 200 Economists and Tech Leaders Warn of A.I. ThreatsAuthor: Ben Casselman Published: July 13, 2026Ben Casselman reports on “We Must Act Now,” a statement warning that artificial intelligence could transform the economy faster than any previous technology and that policymakers need to move faster to understand and respond. The statement says AI may become radically more powerful over the next 10 years, bringing risks such as large-scale job displacement as well as opportunities such as higher living standards. Nearly 200 people signed, including 15 Nobel laureates, the chief economists of OpenAI and Anthropic, Anthropic co-founder Jack Clark, former Google CEO Eric Schmidt, and venture capitalist Vinod Khosla.The killer detail is who joined the warning. Casselman notes that the signatories include economists who have historically been skeptical of Silicon Valley's most dramatic AI job-loss forecasts, including Daron Acemoglu and Simon Johnson, the MIT professors who won the 2024 Nobel in economics. Erik Brynjolfsson, who helped organize the statement, says there has been a notable change in the profession and that economists and policymakers are not ready for the “tsunami” he sees coming.The pull is the measurement problem. The statement does not offer a specific policy menu, but calls for economists, policymakers, and industry leaders to understand the economics of transformative AI and steer it toward complementing humans. Brynjolfsson says one high priority is better data on AI's spread and impact, because current measures tell conflicting stories about job losses and which workers are most exposed.Read more: The New York TimesWhy I Didn't Sign the AI Open LetterAuthor: Andrew McAfee Published: July 13, 2026Andrew McAfee explains why he did not sign “We Must Act Now,” the AI economy statement organized in part by his longtime collaborator Erik Brynjolfsson. McAfee agrees with the letter's starting point that AI is likely to become radically more powerful over the next decade and that it is a general-purpose technology. His objection is not to urgency or to studying AI's economic effects, but to the framing of risk, displacement, and institutional steering as the first move.The killer detail is McAfee's line edit. He says the original letter comes close, then “bounces off the crossbar” by calling for incentives, guardrails, and institutions to steer AI before we know enough about its actual impacts. He points to mixed current evidence: labor-market canaries, but also rising software job postings, low unemployment for younger workers, rising real median income, and claims that AI-adopting companies are adding workers faster than low-adopting peers. His worry is that the letter leans toward upstream governance and dirigisme when the evidence may call for capability building instead.The pull is his replacement statement. McAfee keeps the three-paragraph structure but changes the emphasis: AI is likely to become radically more powerful; like earlier world-changing technologies it will raise living standards while also bringing harms and shocks; and economists, policymakers, and technology leaders should build the capabilities to respond quickly and effectively. It is a concise version of the permissionless-innovation case inside the AI policy debate.Read more: The Geek WayOwn Your WeightsAuthor: Jamin Ball Published: July 10, 2026Jamin Ball argues that the enterprise AI debate about whether companies should “own their weights” or rent models from frontier labs is asking too narrow a question. A model weight file gives a company control over a point-in-time artifact, but not durable control over the capability stack. In his framing, the weight file is a melting ice cube: it does not get worse in absolute terms, but it falls behind as frontier systems improve and enterprise needs change.The killer detail is what Ball says companies really need to own: the data flywheel, reinforcement learning infrastructure, and evaluation harness that produce and improve the model. Simply deploying an open-weights model and declaring sovereignty leaves the enterprise with yesterday's capability and no way to compound workflow-specific learning.The pull is that enterprise AI control may be less about model ownership than operating ownership. The defensible layer is the system that turns company data, edge cases, business definitions, and evaluations into continuously improving performance.Read more: Clouded JudgementWays to Think About Token PricingAuthor: Benedict Evans Published: July 9, 2026Benedict Evans argues that today's AI token prices are a temporary signal from a supply-constrained market, not a reliable guide to long-term value capture. The open question is whether foundation models keep durable pricing power or become commodity infrastructure as data-center capacity, inference efficiency, and model competition all shift. His current read is that the visible market dynamics point toward commoditization unless something materially changes.The killer detail is the mobile data analogy. Evans says cellular networks became a trillion-dollar industry with hundreds of billions in capex after data usage exploded, but carrier stocks went nowhere because value moved up the stack. Tokens may behave similarly: an opaque unit tied to marginal cost, sold through bundles, essential to everything, yet not necessarily where profits accrue.The pull is uncertainty, not prediction. Evans lists paths to model dominance, including network effects, less competition, regulation, export controls, or a lab pulling ahead on execution, but says each requires a new fact not yet visible. Without that change, the model layer looks more like infrastructure beneath the products that capture value.Read more: SourceAlex Karp Is Saying What Every Angry CEO Is Thinking About AIAuthor: Tim Higgins Published: July 11, 2026Tim Higgins reports that Palantir CEO Alex Karp has turned corporate frustration with AI labs into a public argument about enterprise control. Palantir released a white paper, “Institutional Sovereignty in the Age of AI,” laying out steps companies and governments can take to protect themselves from OpenAI, Anthropic, and other foundation-model providers. The article links that paper to Karp's CNBC appearance, where he said “something has gone completely wrong” in the relationship between AI labs and customers and argued that enterprises are paying for tokens that create little value.The killer detail is the value-capture question. Higgins writes that Karp's critique has resonated because AI labs may gain power and insight from customer data, workflows, and decision-making, even when enterprise policies say customer data are not used for training. David Sacks amplified the concern by arguing that Anthropic is moving from the model layer into vertical applications such as science, security, legal, and coding, raising the fear that model providers will watch where value is being created and then move into those markets directly.The pull is that Karp is not alone, even if his style is unusually combative. Higgins notes that Satya Nadella has also warned that companies need to retain the learnings created when they use AI models, while Mark Zuckerberg has framed Meta's new model release partly around lower-cost frontier intelligence. The article presents Karp's campaign as one sign that established technology companies and large enterprises are trying to define where they fit when AI labs become central infrastructure, application competitors, and potential IPO giants at the same time.Read more: The Wall Street JournalThe AI Agents Are Coming for Microsoft OfficeAlex Wilhelm | Cautious Optimism | July 11, 2026Alex Wilhelm argues that one of the week's quieter AI questions is whether the productivity market that Microsoft successfully moved into subscription software is now being attacked by agentic tools. The piece begins with the infrastructure backdrop: SK Hynix raised $26.5 billion in a U.S. listing while building U.S. HBM and advanced-packaging capacity, and memory, chip, and foundry companies are now priced for sustained AI demand.Wilhelm then says the AI conversation has shifted quickly from raw capability to cost per task. He cites new model releases and vendor language emphasizing cheaper agentic and coding models, faster performance, and lower dollars per task. That matters because lower costs make it more plausible for AI systems to take on routine knowledge work at scale rather than remain a premium coding assistant market.The core of the article is Microsoft Office. Wilhelm notes that Microsoft turned Office from a one-time purchase into Microsoft 365, a large recurring revenue business with tens of millions of subscribers and a major productivity segment. Now, he says, late-stage unicorns and AI labs are pushing into the same territory: Anthropic's Cowork was reportedly used mostly outside software development, OpenAI merged ChatGPT and Codex into a tool for creating sheets, slides, docs, web apps, and long-running work, and other companies are building agentic coworkers that connect business data to documents, workflows, schedules, alerts, and apps.The article's caveat is that Microsoft has survived major platform shifts before. The argument is not that Office disappears quickly, but that the definition of office software is broadening from documents and spreadsheets into AI systems that can create, monitor, and act across workplace data.Read moreWhat Is Loop Engineering, and Who Owns It?Author: Nilesh Barla Published: July 11, 2026Nilesh Barla argues that “loop engineering” is becoming a distinct discipline because production AI agents now fail less at single prompts than at runtime: when to stop, what state to preserve, and how to recover after a bad step. Prompt engineering shapes one model call, and context engineering shapes what the model sees, but loop engineering shapes what a sequence of calls actually does.The killer detail is the three-primitives frame. Barla says a real agent loop needs halt conditions, state carryover, and recovery paths, then maps teams across five maturity levels. At the lowest level, an agent is just a model call in a for-loop with a step cap and raw history; by the higher levels, the system has structured state, explicit planning, replay, evaluation, and self-repair.The pull is organizational. If agents are becoming production systems rather than demos, someone has to own the runtime itself. The loop engineer is the role Barla gives to the person responsible for making long-running agent work dependable.Read more: Adaline LabsThe Fight Against AI Data Centers Is Just BeginningEmma Roth | The Verge | July 12, 2026Emma Roth argues that community resistance to data centers has moved from an early warning sign into a national political fight as AI facilities grow larger, more power-hungry, and more visible to nearby residents. The article starts with Apple's failed 2015 plan for a $1 billion data center in Athenry, Ireland, where a small group of residents challenged the project over noise, light pollution, flooding, traffic, and wildlife effects until Apple abandoned it in 2018.The current data-center buildout is presented as much larger and more contentious. Roth writes that residents now cite rising energy costs, water quality, noise, light pollution, and greenhouse gas emissions, while the U.S. Energy Information Administration expects commercial energy demand to surpass residential demand this year because of AI data centers and Goldman Sachs expects data-center power demand to double by 2027.The central evidence comes from Data Center Watch, which says protesters blocked or delayed at least 75 U.S. projects worth $130 billion from January to March, with active opposition groups more than doubling from 396 at the end of 2025 to 833 by the end of the first quarter of 2026. Roth also cites QTS abandoning a $12 billion Wisconsin campus, Delaware City regulators blocking a 580-acre project under the Coastal Zone Act, opposition stopping a QTS project in Prince William County, and pressure that pushed Kevin O'Leary to downsize the proposed 40,000-acre Project Stratos in Utah.The policy section describes a split between federal acceleration and local resistance. President Trump has treated data centers as part of the AI race with China and fast-tracked construction, while some Republican candidates are distancing themselves from that position ahead of midterms. Sanders and Ocasio-Cortez have proposed a moratorium until price and environmental protections exist, bipartisan lawmakers are backing ratepayer-protection measures, and states including Florida, Idaho, and Washington have passed rules on cost shifting, water use, and tax breaks. Roth's caveat is that the policy patchwork is still incomplete, leaving many communities to fight project by project.Read more6 months to live for open modelsAuthor: Nathan Lambert Published: July 12, 2026Nathan Lambert argues that open-weight AI models are facing their most serious policy test so far because U.S. officials are beginning to discuss concrete controls rather than abstract safety concerns. He says reported White House conversations about a new executive order may initially target Chinese-origin models and government use, but could create a broader review habit for frontier open models. His forecast is that a model above the capability range of GPT-5.5, Claude Opus 4.8, or GLM-5.2 could trigger a ban or indefinite delay within six months.The post separates two policy fights that are becoming intertwined: distillation and frontier capability. Lambert says the distillation campaign against Chinese models has become a form of regulatory capture because Anthropic and other closed-model companies would gain economically if Chinese open models were banned. He does not dismiss IP protection, but argues that if a closed model's capabilities are dangerous enough to justify restricting open models, the lab also has to explain why those capabilities are exposed through a queryable API. He cites unauthorized access to Anthropic's Mythos private beta as evidence that APIs are not automatically secure.The broader claim is that a unilateral U.S. ban would hurt positive actors more than bad actors if comparable open models remain available elsewhere. Lambert says the only durable ceiling would require global agreement, which does not exist, and that open models can improve safety by allowing broad inspection, adaptation, and understanding. His proposed near-term off-ramps are a strong U.S. open model release from companies such as Microsoft, Meta, or Reflection, and a broader coalition of open-source beneficiaries lobbying for safe rollout rather than prohibition.Read more: SourceAmericans Deserve a Dividend From AI Companies' RichesAuthor: Scott Stanford Published: July 14, 2026Scott Stanford argues that proposals to give the government a stake in AI companies miss the point unless ordinary citizens directly receive and control the upside. Sam Altman has discussed giving up equity in OpenAI, Washington already owns a stake in Intel, Nvidia is sharing China chip revenue, and Bernie Sanders wants large AI labs to contribute half their stock to a sovereign wealth fund. Stanford says those ideas all park value with the state, not with people.The killer detail is New Carlisle, Indiana, where AWS's Project Rainier is turning cornfields into one of the world's largest AI superclusters. The project is planned to run up to a million chips, draw more than two gigawatts of power, and represents an investment that has grown from $11 billion to $13.8 billion. Stanford uses that local transformation to argue that AI's public bargain should be visible at the household level.The pull is design. A citizen AI dividend would have to specify who earns a stake, how they hold it, and when they see cash. Without that mechanism, the AI wealth debate remains a fight over government balance sheets rather than public ownership.Read more: SourceWho Gets to Define the Frontier?Author: Mark Daley Published: July 14, 2026Mark Daley argues that Demis Hassabis is right to call for a serious institution to verify frontier AI systems, but that the power to test models is also the power to govern them. Hassabis's proposed Frontier AI Standards Body would get privileged pre-release access to advanced models, testing compute, held-out evaluations, support from national labs and security agencies, third-party auditors, and eventually authority to block models from the American market or coordinate a slowdown.The killer detail is Daley's constitutional objection. He says the proposal sometimes looks like a scientific lab, a standards body, an industry regulator, a licensing authority, and an emergency security council at once. Combining those roles because each requires technical expertise would be like putting the central bank, auditor-general, and Supreme Court in one building and calling it efficient.The pull is standard-setting. Daley's concern is not that verification is unnecessary, but that whoever writes the tests, decides what passes, adjudicates disputes, and grants market access may end up defining the frontier itself.Read more: SourceGPT-Red: Unlocking Self-Improvement for RobustnessOpenAI | OpenAI | July 15, 2026OpenAI describes GPT-Red as an internal automated red-teaming model trained to find prompt-injection vulnerabilities at a scale human red teams cannot match. The post says AI systems increasingly encounter third-party data through browsers, connected apps, local files, and tools, creating opportunities for malicious instructions hidden in emails, webpages, tool responses, or code repositories. Human red-teaming remains part of OpenAI's safety process, but the company says it is time-intensive and cannot generate enough diverse adversarial examples for model training.The system is trained through self-play reinforcement learning, with GPT-Red rewarded for eliciting valid failures and defender models rewarded for resisting attacks while still completing their tasks. OpenAI says the training environments specify threat models across settings such as local files, webpage banners, email bodies, and tool outputs. The model is kept separate from deployed production models because it is intentionally trained with malicious capabilities.OpenAI reports that GPT-Red generalized beyond its training set, including an internal replication of the indirect prompt-injection arena from Dziemian et al. (2025), where it found successful attacks in 84% of scenarios compared with 13% for human red-teamers. The post also says GPT-Red transferred attacks from simulation to a live autonomous vending-machine agent, causing price changes and order cancellations, and outperformed a prompted GPT-5.5 baseline against a Codex CLI agent on held-out data-exfiltration tasks.The article's main robustness claim is that OpenAI has used GPT-Red and predecessor models in training since GPT-5.3, with later GPT releases becoming more resistant to prompt injections. It says GPT-5.6 Sol has six times fewer failures on OpenAI's hardest direct prompt-injection benchmark than the best production model from four months earlier, that a “Fake Chain-of-Thought” attack class fell from more than 95% success against GPT-5.1 to below 10% against GPT-5.6 Sol, and that GPT-5.6 Sol fails on only 0.05% of GPT-Red's direct prompt injections. OpenAI says general capabilities and targeted over-refusal evaluations were not harmed, and says a preprint with more details will follow.Read moreAnthropic, Blackstone bet the next trillion-dollar AI business is implementation, not just modelsRebecca Bellan | TechCrunch | July 15, 2026Rebecca Bellan reports that Ode with Anthropic is the $1.5 billion AI implementation company launched by Anthropic with Blackstone, Hellman & Friedman, Goldman Sachs, and other backers. The article says the venture reflects a growing belief among frontier AI labs that enterprise adoption requires more than better models: customers need engineers who can embed inside businesses and turn AI into working systems.Ode was originally conceived by Blackstone after it used both large consulting firms and smaller AI services boutiques across its portfolio companies. TechCrunch reports that Fractional AI, an AI engineering services startup, stood out and was acquired by the joint venture shortly after the venture was announced. Fractional now forms the foundation of Ode, which has 100 engineers and works closely with Anthropic's applied AI team to identify where the technology can affect specific businesses.Ode CEO Chris Taylor tells TechCrunch that the company could someday become a trillion-dollar business if it scales without losing quality. He says an ideal customer is one whose CEO treats the AI project as a top one or two priority, whether it is a major product feature or the reworking of a core business process. Ode will operate under a “Claude-first” principle, using Anthropic technology whenever possible, but the article says it can use rival AI products when needed.The article's central implementation argument comes from Ode chief technologist Eddie Siegel, who says model selection matters but is not where most of the engineering effort goes. He compares it to the choice of programming language in software: one ingredient in a system that still has to be engineered. Bellan writes that Ode's challenge is hiring and training enough elite generalist engineers, many of them former founders, while competing with OpenAI's The Deployment Company and consulting giants that have built their own forward-deployed engineering teams.Read moreVint Cerf is working on a plan to unleash AI agents on the open internetTim Fernholz | TechCrunch | July 15, 2026Tim Fernholz reports that Vint Cerf, after leaving Google, is advising Innovation Labs on an open architecture for identifying AI agents online. Innovation Labs is a subsidiary of Identity Digital, a DNS registry company, and its proposal is to use domain-name infrastructure as part of a system for agent identity, accountability, and auditability. The premise is that agents will need a way to identify themselves if they move beyond proprietary systems and begin interacting across the open internet.The concrete proposal is DNSid, a registry that links an AI agent to an existing internet domain and uses cryptographic proofs to log its registration over time. Innovation Labs says it is trialing the standard with unnamed hyperscalers and identity companies. Cerf frames the problem around authority and accountability: what authority an agent has, where that authority came from, who is accountable for the agent's behavior, how its identity is established, and why anyone should trust it.The article's caveat is that standards are still emerging and agents are more active than static domains. Cerf says the period may be both fascinating and exasperating because the functionality is powerful and interoperability is unresolved. He compares the adoption problem to TCP/IP: competing systems may not work together until users push for functional interoperation. He also says an agentic economy is not inevitable, but that people will try to build it because delegating work to agents will be easier.Read more: TechCrunchxai-org/grok-build, now open sourceAuthor: Simon Willison Published: July 15, 2026Simon Willison argues that xAI's decision to open-source Grok Build is best understood as a trust repair move after a severe privacy failure. The CLI had triggered backlash when users realized that running it in a directory could upload the entire directory to xAI's Google Cloud buckets, including one user's reported SSH keys, password manager database, documents, photos, and videos. xAI disabled the feature, said previously retained coding data would be deleted, and released the code under Apache 2.0.The killer detail is what the codebase reveals. Willison counts 844,530 lines of Rust, only about 3% of which appears vendored, and finds remnants of the upload system still present but disabled: gcs.rs contains Google Cloud upload code, while upload_session_state() now returns a hard-coded session_state_upload_unavailable error. He also notes copied or ported tool implementations from Codex and OpenCode, prompt files, and a terminal Mermaid renderer.The pull is that terminal coding agents are becoming large, intricate software systems in their own right. The privacy failure mattered because these tools operate inside the directories where developers keep their most sensitive work; the open-source release matters because trust now depends on inspecting what an agent can see, send, and do.Read more: SourceThe Pulse: What can we learn from Bun's rapid Rust rewrite with AI?Author: Gergely Orosz and Ivan Klaric Published: July 16, 2026Gergely Orosz and Ivan Klaric argue that Bun's AI-assisted rewrite from Zig to Rust is a practical sign of how software engineering changes when models can take on large, bounded migrations with clear feedback loops. The piece does not treat the rewrite as magic: Jarred Sumner first spent hours turning design judgment into a detailed porting guide, then used adversarial review, parallel agents, compiler errors, and tests to force the work toward correctness.The killer detail is the scale. Bun had 535,496 lines of Zig, 1,448 files, and 22 million monthly downloads, making a conventional rewrite a year-long freeze the team could not justify. Using Fable, Sumner split the work across 64 agents, produced about 6,500 commits, and got the migration done in 11 days at an estimated API cost of $165,000.The pull is economic, not theatrical. If a one- or two-year migration can become an 11-day project, AI coding is not just faster autocomplete; it changes which technical debts are worth paying down.Read more: SourceOrphan risks at the frontier of artificial intelligenceAuthor: Andrew Maynard Published: July 16, 2026Andrew Maynard argues that frontier AI safety frameworks are creating “orphan risks”: harms that companies can see, but do not formally own because they are hard to quantify, do not fit catastrophic-risk thresholds, or fall outside audit-friendly compliance machinery. His target is not existing frontier safety work, but the narrowing effect that happens when private companies decide which risks count as governable.The killer detail is Maynard's contrast between measurable model dangers and threats to value. He points to Meta's three-day Galactica collapse, OpenAI's 2023 board crisis, safety-team departures, and wellbeing litigation as examples of risks that damaged trust, culture, legitimacy, or users without fitting cleanly into conventional model-risk categories. The proposed fix is an orphan-risk register: a public record of risks a company considered and chose not to manage, with reasons.The pull is accountability. Frontier developers' internal scoping choices have become a de facto layer of public governance, so the question is no longer only which risks they manage, but which risks they quietly leave outside the frame.Read more: SourceThe Lab of the Future Should Feel Like a Data CenterLatent.Space with Andy Beam and Rafa Gomez-Bombarelli | Latent.Space | July 16, 2026Latent.Space interviews Lila Sciences CTO Andy Beam and chief science officer for physical sciences Rafa Gomez-Bombarelli about the company's attempt to build an AI-run science factory. The post describes Lila's thesis as treating the lab itself as an “infinite token generator”: if internet data drove the first era of AI scaling, experimentally verified scientific data may be the next scarce training source. Lila is trying to produce that data with robotics, lab instruments, orchestration software, and AI models wired into the wet lab.The central analogy is the lab as data center. Instruments are nodes on a graph, a magnetically levitating transport layer moves materials between them, and experiment scheduling looks like a compute queue. Beam says Lila is not simply an automation company, because the point is not just throughput; it is flexibility, generalization, and experiment capture. The post says Lila has built more than 10 trillion experimentally validated “scientific reasoning tokens,” not internet text or biological sequences.The interview ranges across biology, chemistry, drug discovery, materials science, and the limits of automation. It notes that Lila rebuilt one gas-sorption measurement to run roughly 2,500 times faster, claims its general models can transfer priors from small-molecule chemistry to metal-organic frameworks for carbon capture, and describes model-suggested platinum-group-free electrocatalysts that moved from looking boring or wrong to becoming strong performers. The caveats are physical: experiments have runtimes, biology cannot always be accelerated, chains of thought can be unreliable narrators, and reward hacking becomes more dangerous when a model controls a real lab.Read more: Latent.SpaceWhy AMI Labs' Alexandre LeBrun won't call his AI “AGI” or “superintelligence”Kate Park | TechCrunch | July 16, 2026Kate Park interviews AMI Labs CEO Alexandre LeBrun about why Yann LeCun's world-model startup avoids the language of “AGI” and “superintelligence.” LeBrun says the terms are not useful because they lack stable definitions: “We never used the word AGI. And I just noticed that nobody is using it anymore; they switched to superintelligence.” His argument is that the practical frontier is not a label, but whether AI systems can understand and predict real-world states.The article explains the world-model thesis by contrasting language prediction with physical-state prediction. A large language model predicts the next word; a world model predicts the next state, such as what happens when a glass tips over. LeBrun says LLMs remain complementary and efficient for language, but the physical world is where current AI is weak. Robotics is the clearest case: hardware has advanced quickly, but robots are still brittle outside controlled routines because they lack context and situational understanding.AMI is still pre-product, but TechCrunch reports that LeBrun was in Seoul looking for industrial partners, researchers, and global companies. He says world models cannot be built entirely inside a lab because they need access to real environments. That is why South Korea appeals to AMI: robotics, semiconductors, manufacturing, and fast adoption create the kind of hardware-heavy context that software-only AI has barely touched.Read more: TechCrunchKimi K3 Tech Blog: Open Frontier IntelligenceKimi | Kimi | July 16, 2026Kimi introduces Kimi K3 as an open 3T-class frontier model aimed at coding, knowledge work, reasoning, multimodality, and long-context agentic use. The source describes the model as a 2.8T-parameter system built on Kimi Delta Attention and Attention Residuals, with native multimodality and a 1M-token context window. It says Moonshot AI plans to release model weights by July 27.The post presents K3 through benchmark and use-case sections rather than as a general product announcement. It reports results across coding, productivity, agentic, and multimodal evaluations, including DeepSWE, Terminal-Bench 2.1, Program Bench, SWE Marathon, FrontierSWE, PostTrain Bench, OfficeQA Pro, SpreadsheetBench 2, MCP Atlas, AutomationBench, BrowseComp, GDPval-AA v2, AA-Briefcase, MMMU-Pro, MathVision, BabyVision, OmniDocBench, and PerceptionBench. The source says all reported K3 results use maximum reasoning effort with temperature and top-p set to 1.0, and that different benchmark comparisons use KimiCode, Claude Code, or Codex harnesses depending on the test.Kimi's caveats are unusually concrete. The limitations section says K3 was trained in preserved thinking-history mode, so quality may become unstable if an agent harness does not pass historical thinking content correctly or if an ongoing session switches to K3 midstream. It also says K3's emphasis on long-horizon tasks can make it excessively proactive when it encounters minor issues or ambiguous intent, and recommends imposing explicit behavioral constraints for applications that require strict boundaries. The post adds that K3 remains behind Claude Fable 5 and GPT 5.6 Sol in user experience despite being competitive overall.Read moreVenture CapitalThree Years InAuthor: Tomasz Tunguz Published: July 10, 2026Tomasz Tunguz marks Theory Ventures' third anniversary by arguing that AI's central market effect is time compression. In his telling, model release cycles, company revenue milestones, enterprise adoption, and venture categories have all accelerated. Seed, Series A, and Series B still exist as financing labels, but they no longer cleanly describe company maturity when some seed rounds are larger than IPOs and the best AI companies can mature much earlier than prior software companies.The killer detail is the shift from models to inference. Tunguz argues that inference has become the dominant AI market because workloads and buyer preferences are fragmenting: video, batch, local, agentic, and real-time tasks each create different infrastructure needs. He compares this to databases splitting into OLTP, OLAP, vector, and streaming categories, with AI pushing the same specialization into inference infrastructure.The pull is that Theory sees the AI-native venture firm as part of the same pattern. The firm says it has analyzed twice as many investment opportunities with three investors working alongside a nine-person intelligence organization, using agents and research systems to map markets, source companies, and support diligence. The piece is both a market map and a statement about how venture itself is being rebuilt by the technology it funds.Read more: LinkedInVenture Has Rarely Looked More BifurcatedAuthor: Beezer Clarkson Published: July 14, 2026Beezer Clarkson points to PitchBook's Q2 report as evidence that the U.S. venture market has split into two very different realities. AI now accounts for more than 60 percent of all U.S. venture deal value, meaning the headline market can look active and well-funded even while much of the non-AI market is dealing with a much colder liquidity and fundraising environment.The thread uses that split as the setup for Clarkson's latest Origins episode with Alec Litowitz, founder of Magnetar and QStar Capital and one of Citadel's original founding partners. Clarkson says markets like this are periods of genuine uncertainty, not merely ordinary risk, which is why Litowitz's Adaptability Quotient framework is relevant.The embedded clip makes the liquidity point concrete. Litowitz says DPI is “the resolution of uncertainty” because it converts an uncertain investment into actual cash returned to LPs. In his framing, a realized dollar is a real mark, while TVPI remains uncertain until it is realized.The killer detail is the distinction between pricing risk and resolving uncertainty. Litowitz's perspective matters because QStar is a SpaceX investor and Clarkson says the conversation happened just before one of venture's most consequential IPOs. The episode's stated questions are why venture remains a way to gain exposure to innovation, how AI is changing what is investable, why liquidity is ultimately a function of time, and why uncertainty requires a different decision framework from risk.Read more: XThe Best Angel Investors in the US: Who Backs the Most Unicorns, and Who's Active NowAuthor: Ilya Strebulaev Published: July 10, 2026Ilya Strebulaev ranks angels, angel groups, accelerators, and incubators by lifetime U.S. unicorn investments, counting checks written before a company reached unicorn status. The top of the combined list is dominated by organizations: Y Combinator leads with 113 unicorn investments, followed by Plug and Play at 52 and 500 Global at 41. Sand Hill Angels is the highest-ranked angel group at 31.The killer detail is how quickly the list changes below the biggest accelerators. Strebulaev says 271 of the 304 investors in the Top 200 are individuals, or 89%. In the top 100, individuals are 91%. That makes the market underneath the large accelerator counts look much more personal: mostly operators and individual angels writing early checks from their own networks.The pull is the ranking's own caveat. Strebulaev writes that every lifetime leaderboard has a blind spot because many of the unicorns behind those totals were founded a decade or more ago, and some angels have since moved into formal funds, slowed down, or stopped investing. His post therefore separates lifetime performance from recent cohorts, including companies founded in 2015 or later and 2020 or later. For founders or allocators making current decisions, that distinction matters: a career record and a current record are not the same measure.Read more: Ilya StrebulaevAre Prediction Markets Doomed to Fail?Author: Contrary Published: July 16, 2026Contrary argues that prediction markets' current boom depends on whether platforms can prove they are more than regulated gambling with exchange-style branding. Kalshi and Polymarket have reached mass cultural, investor, and regulatory attention, but the article says the underlying idea is old: academic markets, corporate forecasting tools, Intrade, PredictIt, and other predecessors all struggled with the same linked problems of liquidity, legality, and user appeal.The killer detail is the comparison with sportsbooks. Prediction markets present themselves as peer-to-peer, transparent, and non-house-based, but sports contracts reportedly account for more than 90 percent of Kalshi trading, and the article says the platforms keep a much thinner slice of volume than sportsbooks. A market can therefore show sports-betting-scale handle while generating far less revenue.The pull is that the product's hardest problem may be distribution of wins. If a small group of sharp traders captures most profits while casual users lose interest, prediction markets may become valuable data feeds and professional tools before they become durable consumer networks.Read more: SourceRegulationExclusive: The Next Frontier of the Deportation Wars: College CampusesAuthor: Adrian Carrasquillo Published: July 11, 2026Adrian Carrasquillo reports that college campuses are becoming a new front in the fight over immigration enforcement because automatic license plate readers can turn ordinary campus security infrastructure into searchable location data. His thesis is that Flock Safety's camera network, even without direct ICE or DHS contracts, can feed deportation enforcement through local police partnerships and data-sharing practices.The killer detail is the campaign target. The Emergency Campaign to Support Higher Education, working with Schools Drop ICE, is focusing on 75 colleges and universities publicly identified as having Flock contracts. Flock says it has no ICE or DHS contracts, but activists argue the risk comes through local agencies that coordinate with federal authorities and run searches on their behalf.The pull is broader than immigration. Carrasquillo notes that license plate readers have already been abused by officers for stalking, and that Flock's AI search features can identify more than plates, including bumper stickers. A campus safety tool can become a political surveillance system when the data layer is searchable.Read more: The BulwarkThe Supreme Court Broke Independent Agencies. Here's a Way to Slow the Damage.Author: Todd Phillips Published: July 12, 2026Todd Phillips argues that the Supreme Court's decision in Trump v. Slaughter damaged independent agencies by ending for-cause removal protections, but did not leave Congress powerless. The ruling weakens the old model in which commissioners at bodies such as the FTC, NLRB, CPSC, SEC, and CFTC could be insulated from dismissal over policy disagreements. Phillips says the next fight is whether presidents can turn nominally bipartisan commissions into one-party instruments.The killer detail is the procedural fix: quorum rules. Phillips proposes that Congress require bipartisan slates of commissioners to be seated before independent agencies can act. A president

Get the 200+ Page Optimal Living Daily Workbook (PDF) — Free. Want to turn today's episode into an actionable plan? Join the Optimal Living Weekly newsletter and I'll send you our 200-page digital workbook immediately. It's packed with the best takeaways from the show, formatted for easy reading and implementation at home. Get your free PDF workbook here: ⁠⁠https://oldpodcast.eo.page/join⁠ Discover all of the podcasts in our network, search for specific episodes, get the Optimal Living Daily workbook, and learn more at: OLDPodcast.com. Episode 4086: Shirley shares a practical framework for cutting through overwhelm by identifying what matters most, then aligning your relationships, thoughts, and daily decisions with those priorities. Her simple exercises can help you gain clarity, reduce mental clutter, and create more space for the things that truly matter in this season of your life. Read along with the original article(s) here: https://daringliving.com/how-to-simplify-focus-in-your-life/ Quotes to ponder: "Know what's important to you and use that as your focus in this phase in your life." "Everything is your decision, and what you choose to think about it." "Instead of trying to change others or make them see you a certain way, what if you can simplify the relationship to building genuine connections and be interested to learn about them and hear what others have to say?" Episode references: Mark Zuckerberg: https://en.wikipedia.org/wiki/Mark_Zuckerberg Steve Jobs: https://en.wikipedia.org/wiki/Steve_Jobs Learn more about your ad choices. Visit megaphone.fm/adchoices

Podcasting is lousy with guests who peddle advice about starting businesses, building careers, and living full lives. Few of these gurus have actually accomplished anything. Fewer still have advice that comes anywhere close to revelatory. Mark Pincus is the rare exception. The founder of 10 companies, Mark is best known for Zynga, the gaming juggernaut behind FarmVille and Words With Friends. At its peak, Zynga was valued at more than $12 billion and accounted for 20% of Facebook's page views. It was so big, in fact, that Mark Zuckerberg once admitted to Mark Pincus, “Zynga is the only company that is capable of being an actual Facebook competitor.” Across his 30-year career, Mark — Pincus, not Zuck — developed a cheat code for building products people love and living a life anyone would admire, and that cheat code winds its way through his new book, Life at the Speed of Play. Today, he shares how he learned to trust his gut, why most founders build too much and test too little, and how failure taught him to move faster, pivot sooner, and search for real signals instead of hope.

Here they come again: Billionaires, wailing that Congress must – MUST! – act immediately to slash the monthly Social Security checks that middle-class and poor retirees count on.Plutocratic elites and their anti-government ideologues periodically erupt in outrage that elderly Americans who've earned retirement benefits are depleting the Social Security Trust Fund. So, they exclaim, government must cut the payments these old folks are getting.But wait – it's not “the government's money.” It belongs to the retirees themselves. They've paid monthly payroll taxes into the fund for years on the guarantee that they would later draw benefits out.Maybe so, bark opponents, but the money well is going dry, so the only way to “save” the program is to chop payments owed to beneficiaries.In three words: That's a lie.What the superwealthy don't want us to notice is that the Social Security tax is spectacularly unfair. If your yearly income is less than $185,000 (which includes 95% of us) – every penny of your earnings is subject to retirement tax. But if you're paid a million a year, or a billion, or even more – everything over $185,000 is tax free. Sweet!Wait, there's more. Instead of being paid wages, most über-wealthy people draw their annual income from a Wall Street scheme called “unrealized capital gains.” Big surprise – those gains are totally exempted from our nation's retirement tax.This is Jim Hightower saying… So, let's make Musk, Zuckerberg, Bezos, and other tax-dodging billionaires pay on all of their income like the rest of us do. That's only fair. Then America can increase benefits so everyone can have a dignified retirement. Now that's true fairness!Do something!To get involved with the fight to make sure Social Security and other social safety net programs stay strong, check out Social Security Works at socialsecurityworks.org.Jim Hightower's Lowdown is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit jimhightower.substack.com/subscribe

AGENDA: 00:00 – Apple SUES OpenAI: Did They Steal Apple's Biggest Secrets? 05:10 – Is OpenAI's $6BN Hardware Bet Already Dead? 12:50 – Zuckerberg Is Back: Meta Finally Takes On OpenAI 18:05 – The AI Spending Bubble Nobody Is Talking About 23:45 – Claude Is Coming for Designers, Product Managers & Figma 27:15 – Anthropic's $50BN Explosion: Have We Already Hit AI's TAM? 36:00 – The $26BN AI IPO Powering the Entire Industry 40:00 – Seed Investing Is Dead? Jason Calacanis Changes Strategy 57:00 – SaaS Is in Trouble: AI Is Accelerating Terminal Decay 01:15:00 – Why Greylock Said No to Billions of Extra Dollars  

Get the 200+ Page Optimal Living Daily Workbook (PDF) — Free. Want to turn today's episode into an actionable plan? Join the Optimal Living Weekly newsletter and I'll send you our 200-page digital workbook immediately. It's packed with the best takeaways from the show, formatted for easy reading and implementation at home. Get your free PDF workbook here: ⁠⁠https://oldpodcast.eo.page/join⁠ Discover all of the podcasts in our network, search for specific episodes, get the Optimal Living Daily workbook, and learn more at: OLDPodcast.com. Episode 4086: Shirley shares a practical framework for cutting through overwhelm by identifying what matters most, then aligning your relationships, thoughts, and daily decisions with those priorities. Her simple exercises can help you gain clarity, reduce mental clutter, and create more space for the things that truly matter in this season of your life. Read along with the original article(s) here: https://daringliving.com/how-to-simplify-focus-in-your-life/ Quotes to ponder: "Know what's important to you and use that as your focus in this phase in your life." "Everything is your decision, and what you choose to think about it." "Instead of trying to change others or make them see you a certain way, what if you can simplify the relationship to building genuine connections and be interested to learn about them and hear what others have to say?" Episode references: Mark Zuckerberg: https://en.wikipedia.org/wiki/Mark_Zuckerberg Steve Jobs: https://en.wikipedia.org/wiki/Steve_Jobs Learn more about your ad choices. Visit megaphone.fm/adchoices

Get the 200+ Page Optimal Living Daily Workbook (PDF) — Free. Want to turn today's episode into an actionable plan? Join the Optimal Living Weekly newsletter and I'll send you our 200-page digital workbook immediately. It's packed with the best takeaways from the show, formatted for easy reading and implementation at home. Get your free PDF workbook here: ⁠⁠https://oldpodcast.eo.page/join⁠ Discover all of the podcasts in our network, search for specific episodes, get the Optimal Living Daily workbook, and learn more at: OLDPodcast.com. Episode 4086: Shirley shares a practical framework for cutting through overwhelm by identifying what matters most, then aligning your relationships, thoughts, and daily decisions with those priorities. Her simple exercises can help you gain clarity, reduce mental clutter, and create more space for the things that truly matter in this season of your life. Read along with the original article(s) here: https://daringliving.com/how-to-simplify-focus-in-your-life/ Quotes to ponder: "Know what's important to you and use that as your focus in this phase in your life." "Everything is your decision, and what you choose to think about it." "Instead of trying to change others or make them see you a certain way, what if you can simplify the relationship to building genuine connections and be interested to learn about them and hear what others have to say?" Episode references: Mark Zuckerberg: https://en.wikipedia.org/wiki/Mark_Zuckerberg Steve Jobs: https://en.wikipedia.org/wiki/Steve_Jobs Learn more about your ad choices. Visit megaphone.fm/adchoices

When Mark Lanier and his young client Kaley faced Meta and Google in an LA courtroom earlier this year, it seemed a bigger battle than David v Goliath. Lanier, however, was determined to prove the companies had not just stumbled into a youth mental health crisis, but had helped to engineer it. Help support our independent journalism at theguardian.com/infocus

David Stout is the founder of WebAI, an artificial intelligence company publicly valued at $2.5 billion. After growing up on a Michigan ranch, he launched a web services company in college that generated more than $750,000 annually before leaving school to pursue AI.In this episode, David explains how a $400 loan, repeated investor rejection, and a relentless focus on on-device intelligence shaped WebAI's rise. He also discusses AI ownership, data centers, leadership, faith, recruiting elite talent, and the future of specialized artificial intelligence.Hosted on Ausha. See ausha.co/privacy-policy for more information.