Podcasts about watermarking

In this episode of Business Brain, we open with a management tip that could be costing you right now: tell your staff to use as much AI as they need. You’ll hear why employees quietly ration their own usage out of misplaced budget-consciousness, why a-la-carte tokens still beat leveling up plans you don’t need, and how a quick conversation about limits can surface whether someone’s underusing a tool that could transform their work. Teach your team never to assume a tool is too expensive without asking you first. Then we tackle the AI watermarking debate, sparked by an EU mandate and the first company to publicly embrace it. You’ll learn how text watermarking actually works through word choice rather than hidden characters, why it isn’t deterministic, and what happens when you run watermarked output through another model. We spar over regulatory capture, the AI triopoly, and whether “doing it right” is the brand or the reality, then point you to the tools people are already building to strip marks out. It’s a lively, skeptical look at where this is all heading, all in the spirit of the charmed life. 00:00:00 Business Brain – The Entrepreneurs' Podcast #783 for Casual FridAI, August 28th, 2026 August 28th: National Bow Tie Day 00:01:37 Tell your staff to use as much AI as they need Loop Engineering 101 00:07:54 SPONSOR: Hims. With Wegovy® at Hims, lose up to 20% of your body weight when combined with diet and exercise. Visit https://hims.com/businessbrain to get a personalized, affordable plan that gets you. 00:09:29 SPONSOR: SomniPods 3 sleep earbuds from Fitnexa: Flat enough to sleep on, and the app is the real prize: https://go.fitnexa.com/brain automatically applies $10 off (or use code BRAIN). 00:11:02 Watermarking on other models 00:17:59 How Watermarking Works 00:20:55 Watermarks-Remover Script 00:23:41 Business Brain 783 Outtro This Episode's Big Takeway: Teach your staff not to assume that any tool is too expensive Check out Business Brain Blueprints Tell Your Friends! Business Blueprints Review Business Brain Subscribe to the show feedback@businessbrain.show Call/Text: (567) 274-6977 X/Twitter: @ShannonJean & @DaveHamilton, & @BizBrainShow LinkedIn: Shannon Jean, Dave Hamilton, & Business Brain Facebook: Dave Hamilton, Shannon Jean, & Business Brain The post FridAI – AI Usage and Watermarking – Business Brain 783 appeared first on Business Brain - The Entrepreneurs' Podcast.

In this episode, we provide insights into Anthropic's watermarking breach and its significance in AI security. Additionally, we explore the implications of Google's $12 billion investment in chips. Show LinksGet the top 80+ AI Models for $8.99 at AI Box: ⁠⁠https://aibox.aiHow I Grow and Scale My Business with AI: https://www.skool.com/aihustleGet the AI Chat Daily Newsletter: https://www.aichatdaily.com/newsletter

Apple's camera-equipped AirPods are seen in leaked footage, are they a bad idea? Anthropic reveals how Claude watermarks text (it's not great), Robin Williams' children return to Instagram to fight AI Slop, the college tech load out, and Jason revives Aperture.Member Promo Code: IWANTCHAPTERS (Click above and the $2.50 promo will be auto applied!)Our AppsElement Photo AppKan Do Task AppContextly NotesStephen's CommunityStash Beta AccessSponsors:Claude AI - Ready to tackle bigger problems? Sign up for Claude todayat: claude.ai/primaryOur LinksTop Five Tech | Stephen's PodcastCreative Effort | Jason's PodcastWatch on YouTube!Show Notes via EmailEmail Us: podcast@primarytech.fm@stephenrobles on Threads@jasonaten on ThreadsLinks from the showThe HÅG Capisco Is A Great Chair For People Who Sit Weird - AftermathmacOS 27 Golden Gate beta 6 features redesigned traffic light window controls - 9to5MacApple's Camera-Equipped AirPods Confirmed: See Them in Action - MacRumorsCamera-equipped AirPods reportedly won't launch in 2026, despite demo video leak - 9to5MacApple Accidentally Leaked More Than 10 New Products in macOS Update - MacRumorsApple Maps opens ad booking for businesses, promotional offer available - 9to5MacHow Claude's text watermarking works - AnthropicDaring Fireball: Anthropic's ‘Watermark' Text Adulteration in Claude Is a Perversion of WritingMastodon post from Stephen Robles: "Hey Claude, never say 'opine' again, thx"Back to School Tech Deals 2026declaude: rewrite AI-flavored text as plain proseOpenAI lays out new security changes after its AI hacked Hugging Face | The VergeApple announces changes for apps in the European Union - AppleDisney's ABC Sues FCC Over Challenge to Its Broadcast Licenses - WSJComcast is putting motion sensing into millions of homes — using its existing Wi-Fi routers | The VergeRobin Williams' Kids Revive Instagram to Combat 'AI Abuse' of His Image - TheWrapAn update to how we count public views across YouTube - YouTube CommunityNomad 65W USB-C AC AdapterGoogle Pixelsnap Charger with Stand - Qi2 Wireless ChargerEF ECOFLOW Rapid Pro Power Bank, 4-Port with 300W Max OutputAnker Power Bank (10K, Fusion, Built-In Cable) - Anker USMainstays 3 Drawer Fabric Nightstand with USB, Arctic White Boucle - Walmart.com (00:00) - Intro (02:57) - macOS 27 Buttons (03:55) - AirPods with Camera (16:04) - Apple Maps Ads (16:37) - Sponsor: Claude (19:26) - Claude Watermarking (37:18) - OpenAI Puts on Brakes (38:21) - Apple X EU (41:07) - Disney Sues FCC (44:57) - Comcast Wi-Fi Motion (46:42) - Robin Williams AI Slop (01:06:18) - YouTube Views (01:10:50) - Zuck is Syndrome (01:11:43) - College Tech Load Out (01:17:40) - Element Photo App ★ Support this podcast ★

Cathy Hackl and Lee Kebler break down the biggest shifts reshaping tech and content creation. Explore AI watermarking's role in combating deepfakes, how social media algorithms now reward authenticity over virality, and why Apple's upcoming launches matter for creators. Plus: the creator economy's new degree programs, ethical questions around tipping robots, and what smart glasses mean for privacy. This episode cuts through the hype to reveal what actually impacts creators, brands, and the future of digital culture.Keywords: AI watermarking, social media algorithms, content creation, generative AI, authenticity, privacy, Apple launches, creator economy, smart glasses, ethical technology Sound Bites"Watermarking in AI is a complex challenge.""Upcoming Apple devices could redefine user experience.""Tipping robots raises new ethical questions."Chapters00:00Introduction to the episode and guest01:56The integration of AI into schoolwork and watermarking04:05Watermarking in music and generative AI05:49Challenges of watermarking in AI generated content08:00Recognizing AI-generated content and voice10:02Social media algorithms and human authenticity11:55The return of MySpace and social media evolution14:00Monetization and creator economy updates16:04Upcoming tech launches from Apple and new devices17:59Legal issues surrounding Meta and AI20:08AI in youth and safety concerns22:04The future of smart glasses and XR24:04Apple's new products and privacy focus25:59Parental control and screen time management27:55Robotic bartenders and tipping etiquette30:00Summary and closing remarks

The biggest tech news & social media trends on the internet from August 19th, 2026.Timestamps:00:00 Intro1:43 Jackie Aina investigates trolls and online hate 9:32 Twitch streams are being used to train Amazon AI19:08 Anthropic introduces watermarking for ClaudePlease consider buying us a coffee or subscribing to a membership to help keep Centennial World's weekly podcasts going! Every single dollar goes back into this business

Claude will now watermark the content it generates, and the fight over what that means for writing, schoolwork, and "AI plagiarism" is only starting. Paul and Mike unpack the watermarking news, then move through AI's environmental reckoning, OpenAI's very public friction with the White House, Anthropic's march toward a record IPO, Grok 4.6's return to the frontier, and an AI agent that hacked a gym website while trying to book a class. AI-Pulse Survey: Fill out this week's AI-Pulse Survey here. Show Notes: Access the show notes and show links here Timestamps: 00:00:00 — Intro 00:07:04 — Claude Will Now Start Marking AI-Generated Content 00:22:41 — AI's Environmental Reckoning 00:35:58 — Drama from the White House Over OpenAI Hire 00:45:25 — Anthropic's Hidden Advisor 00:50:04 — Sanders Threatens an AI Pause 00:55:30 — xAI Releases Grok 4.6 00:58:30 — More on Google's AI Leadership Reshuffle 01:02:54 — Revisiting the Responsible AI Manifesto 01:08:09 — AI Agent Hacks a Gym Website 01:12:10 — AI Use Case Spotlight 01:19:14 — AI Product and Funding Updates This week's episode is brought to you by MAICON, our 6th annual Marketing AI Conference, happening in Cleveland, Oct. 13-15. The code POD100 saves $100 on all pass types. For more information on MAICON and to register for this year's conference, visit www.MAICON.ai. Visit our website Receive our weekly newsletter Join our community: Slack Community LinkedIn Twitter Instagram Facebook YouTube Looking for content and resources? Register for a free webinar Come to our next Marketing AI Conference Enroll in our AI Academy 

-Several departures have led to concerns that the company is ignoring safety in favor of growth -As first reported by the Washington Post, the woman, named as Jane Doe 4 in the lawsuit, said that her stepfather used Grok to create "thousands of sexually explicit images of her as a child and traded them online." -The company said its text watermarking will not have easy-to-see visuals, will not be distinguishable to the people who read it and will not be adding hidden characters to the text. Learn more about your ad choices. Visit podcastchoices.com/adchoices

Why Watermark?Claude wants everybody to know it is there. Keith Teare argues that this is exactly the wrong instinct. The issue is not whether AI touched a piece of work. The issue is whether the work is true, useful, accountable, and human-directed.Watermarking starts from suspicionA watermark assumes there is a problem to solve. Keith's objection is that Anthropic appears to be accepting the premise that AI use tarnishes the user. If Claude helped draft, structure, summarize, or edit human-origin work, why is that the fact that must be marked? The mark answers the wrong question: not whether the work is good, true, accountable, or human-directed, but whether Claude was there.Detection tools confuse use with authorshipAndrew ran Keith's editorial through an AI detector and got an 86% AI score; Keith ran Saul Klein's article through one and got 100%. Neither result proves much. Andrew also admitted he uses Claude as a draft and then rewrites it. Keith described the same process: give AI the source material, ask it to surface themes, debate the frame, then rewrite. The question is not whether AI helped. The question is who is responsible for the final work.The shame belongs in the wrong placeKeith accepts that shame has a place: spam factories, fake authorship, unreviewed AI output, fake evidence, and publishing something you cannot stand behind. He feels no shame in "recruiting an army of AI agents" to help accomplish his goals, provided he reads, changes, edits, and owns the output. Society is trying to attach shame to the tool itself. That is the mistake.AI should be cheap and everywhereThe show turns from watermarking to access. Grok Bot at $200 a month is a sign of capability, but also a sign of scarcity. Keith argues that equality in AI is about price and reach: how cheap is it, and how widespread is it? If intelligence is valuable, the goal should be to make it free or nearly free for many use cases, not to add friction that helps the Luddite argument.Abundance requires massive investmentKeith defends the scale of AI investment because demand still exceeds supply. Nvidia's half-trillion-dollar commitment, hyperscaler deals, IPOs, and AI infrastructure financing are not just market exuberance. They are the precondition for AI to reach everyone. As Keith puts it, if you want AI in the hands of an African school child, you have to build the infrastructure first. The risk is not that too much is being built. It may be that too little is being built, or that access is captured at the price and distribution layer.Bottom line: AI is good. Build enough of it for everyone. This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.thatwastheweek.com/subscribe

Stay informed on current events, visit www.NaturalNews.com  - Introduction and Upcoming Content (0:10) - Protecting Genetic Integrity (1:09) - Navy Personnel Welfare Concerns (47:28) - Political Analysis of MAGA Movement (47:48) - AI Warfare and Data Collection (71:43) - AI Regulation and Watermarking (78:28) - Local AI Model Training (83:16) - Biosecurity and Genetic Risks (86:55) - Robotics and Autonomous Systems (97:21) - AI-Designed Pathogenic Proteins (103:24) Watch more independent videos at http://www.brighteon.com/channel/hrreport  ▶️ Support our mission by shopping at the Health Ranger Store - https://www.healthrangerstore.com ▶️ Check out exclusive deals and special offers at https://rangerdeals.com ▶️ Sign up for our newsletter to stay informed: https://www.naturalnews.com/Readerregistration.html Watch more exclusive videos here:

In this episode of Business Brain, we dig into agents and loops — and how most of us are already using them without realizing it. Every time you interact with a chatbot, it acts as an ephemeral agent; the real leverage comes from building persistent ones. Shannon walks through the three-layer system he handed ten grand to inside Claude Cowork: dumb, fixed-rule loops that check prices daily, a weekly management agent, and a third that reports to him with approve-now, veto-now, or auto-execute options. The lesson that keeps surfacing is to keep the LLM out of the daily rules so the whole thing stays unemotional and disciplined. We also share a quick tip: when you can’t find an old chat, let your AI search its own history instead of fighting clunky built-in search. Then we get into Anthropic’s watermarking announcement, where content produced by Claude will carry hidden signals identifying it as AI-generated. We wrestle with the gray area it creates: if we feed our own transcript to Claude for a summary, that’s our content shaped by AI, not conjured from nothing — so the watermark shouldn’t imply Claude created it from scratch. We argue the definition of AI-generated content needs to be far less binary, favoring transparency about the whole chain rather than just the last step, and we look at how open-source models and Apple might reshape all of it. Owning agents, staying curious, and keeping it honest — that’s how we keep living the Charmed Life. 00:00:00 Business Brain – The Entrepreneurs' Podcast #779 for Casual FridAI, August 14th, 2026 August 14th: Navajo Code Talkers Day 00:01:57 Ask your LLM chat to search itself for old chats you can't find. 00:03:47 Agents & Loops 00:12:20 Scribe. Don’t get caught being the only person who knows how something works. For a limited time, book a demo at https://scribe.how/brain and mention BRAIN for your first month of Scribe Capture free. 00:14:20 Claude is embedding watermarks into your text (and everything else) 00:28:40 Business Brain 779 Outtro This Episode's Big Takeway: Learn to create persistent agents and loops Check out Business Brain Blueprints Tell Your Friends! Business Blueprints Review Business Brain Subscribe to the show feedback@businessbrain.show Call/Text: (567) 274-6977 X/Twitter: @ShannonJean & @DaveHamilton, & @BizBrainShow LinkedIn: Shannon Jean, Dave Hamilton, & Business Brain Facebook: Dave Hamilton, Shannon Jean, & Business Brain The post FridAI – Agents + Loops and AI Watermarking – Business Brain 779 appeared first on Business Brain - The Entrepreneurs' Podcast.

The SEC stopped waiting for Congress. Harmony minted four billion tokens out of thin air. And on Robinhood's new chain, AI agents moved $200 million while the humans logged off. Joel and Travis cover the SEC's August 14 vote on Regulation Crypto — the first formal crypto rulemaking of Chairman Paul Atkins' tenure, landing days after the Senate left town without moving the CLARITY Act. Then: Harmony's empty-block exploit that minted 26% of ONE's total supply and the chain rollback the team is now weighing; Goldman Sachs buying NEOS Investments for $2.25 billion and inheriting a bitcoin income ETF; Tether becoming the 17th largest holder of US Treasuries on Earth; Anthropic watermarking everything Claude writes to satisfy the EU AI Act; H100 Group's world-first bitcoin-for-bitcoin acquisition; Hyperliquid's absurd $106 million of revenue per employee; and Anthropic's $9.1 billion, 20-year lease with bitcoin miner Riot Platforms — signed at a moment when it costs more to mine a bitcoin than a bitcoin is worth. Joel demos three AI builds including an America Online time capsule frozen in August 1996, and Travis walks through FourthWeb's agent swarm scraping 150-plus news sources every fifteen minutes. Plus: 52% of Gen Z investors have moved money earmarked for investing into sports betting, and 26% now call it part of their long-term financial strategy. Programming note — this is our second-to-last show before a hiatus. Joel's getting married. We'll be back in late November. We're not quitting. Not financial advice. Stay bad.Support the show: https://badcryptopodcast.comSee omnystudio.com/listener for privacy information.

Lenovo's ThinkBook patent codenamed Project Swan will bring horizontal rollable displays to its laptop form factor, and Apple is reportedly splitting the release of its premium iPhone 18 devices from its cheaper ones.Starring Jason Howell and Tom Merritt.Links to stories discussed in this episode can be found here. Hosted on Acast. See acast.com/privacy for more information.

Anthropic said new Claude models will watermark generated text to satisfy the EU AI Act, and it's out courting investors for a possibly record-breaking IPO. Apple's glass iPhone stayed on track, YouTube doubled its monetization bar, and FlightAware sued Kalshi. Links Anthropic says new Claude models will embed watermarks in generated text and C2PA metadata in files to comply with the EU AI Act, and it will update past models (The Register) Sources: Anthropic is courting investors for what could be the biggest IPO yet, touting rapid growth and plans to address mounting public backlash against AI (The Wall Street Journal) Sources: Apple remains on track to launch a glass-centric design overhaul of the iPhone Pro line in 2027, countering rumors that led Jefferies to downgrade AAPL (Bloomberg) YouTube says that from February 1, new creators will need double, or 8,000, watch hours over the past year or 20M Shorts views in the past 90 days to earn money (TechCrunch) Zuckerberg's long essay returns to his "open" AI arguments at an opportune time, as Chinese open-weight models are "close enough" to frontier at much less cost (Spyglass) Flight tracking platform FlightAware sues Kalshi in New York, alleging Kalshi is using its data without permission to let users bet on flight cancellations (The Wall Street Journal) Subscribe to the ad-free feed.

Google Tests AI-Focused Homepage Design, Ninth Circuit Rejects Meta's Appeal in Youth Mental Health Trial, and Nvidia Launches $500 Billion AI Compute Financing Initiative. MP3 Please SUBSCRIBE HERE for free or get DTNS shows ad-free. A special thanks to all our supporters–without you, none of this would be possible. If you enjoy what you seeContinue reading "Anthropic Adopts AI Watermarking for EU Compliance – DTH"

In this episode, we discuss recent acquisition talks with the robotics firm Physical Intelligence and how this impacts the competition between OpenAI and Anthropic. We also explore Meta's new AI watermarking tool, the launch of Synthesia's roleplay sessions, and the implications of tech companies signing agreements to cover AI-related electricity costs.Chapters00:00 Introduction00:08 Samsung's Robotics Acquisitions00:25 Meta's AI Watermarking Tool00:39 Synthesia's Roleplay Sessions00:55 Mistral's Valuation Surge10:14 AI Electricity Cost Pledge Show LinksGet the top 80+ AI Models for $8.99 at AI Box: ⁠⁠https://aibox.aiHow I Grow and Scale My Business with AI: https://www.skool.com/aihustleGet the AI Chat Daily Newsletter: https://www.aichatdaily.com/newsletter

In this episode, we discuss the implications of Anthropic's exploration of robotics acquisitions. We also examine Meta's developments in AI watermarking technology.Chapters00:00 Introduction00:08 Samsung's Robotics Acquisitions00:25 Meta's AI Watermarking Tool00:39 Synthesia's Roleplay Sessions00:55 Mistral's Valuation Surge10:14 AI Electricity Cost Pledge Show LinksGet the top 80+ AI Models for $8.99 at AI Box: ⁠⁠https://aibox.aiHow I Grow and Scale My Business with AI: https://www.skool.com/aihustleGet the AI Chat Daily Newsletter: https://www.aichatdaily.com/newsletter See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

In this episode, we analyze Anthropic's discussions about robotics acquisitions and what it means for the industry. We also highlight Meta's innovative approach to AI watermarking.Chapters00:00 Introduction00:08 Samsung's Robotics Acquisitions00:25 Meta's AI Watermarking Tool00:39 Synthesia's Roleplay Sessions00:55 Mistral's Valuation Surge10:14 AI Electricity Cost Pledge Show LinksGet the top 80+ AI Models for $8.99 at AI Box: ⁠⁠https://aibox.aiHow I Grow and Scale My Business with AI: https://www.skool.com/aihustleGet the AI Chat Daily Newsletter: https://www.aichatdaily.com/newsletter

We're announcing AIEWF speakers this week! Take the AI Engineering Survey!Today's guest Ethan first joined us for the LS Paper Club as the lead on NVIDIA Cosmos World Model, but then joined xAI and built Grok Imagine in 3 months:He comes back on Latent Space with some nuclear hot takes: that Video Models primarily get their intelligence from LLMs, not from training on video data, and that the next frontier for truly interactive, realtime, long-horizon world models is to work on LLMs (perhaps Interaction Models as well…)Put it this way: In the near term, the next Sora won't be a better video model, but a video agent.Generative Media may more closely follow the evolution of AI coding which went from focusing on one-shot output performance and cost, to multiturn reasoning and planning models for agents and systems that can plan, edit, test, debug, and submit PRs.At a certain point, coding models got so good that the only significant next step to improve performance was handling the orchestration of these models.Now as the performance of video models increases significantly across realism, consistency, & prompt adherence while becoming more cost efficient, the next evolution of video generation may also be systems that can plan, generate, edit, critique, and iterate across an entire creative task. In this episode, Ethan joins swyx and Vibhu to unpack what it actually takes to build frontier image and video systems: data, VAEs, diffusion transformers, audio-video alignment, inference speedups, and the hidden cost of storing and moving massive video datasets. From building NVIDIA's Cosmos world model to joining xAI as Grok Imagine was being built from zero to one, Ethan He has been at the center of some of the most important work in video generation, multimodal models, and real-time world models.We go deep on Grok Imagine, how a small xAI team shipped its first multimodal video model in three months, why iteration speed matters more than almost anything in model development, and why many of the biggest gains come from fixing tiny bugs in data and training pipelines. Flipbook: The future of VideomaxxingVideo agents are almost a sure bet to be the trend in the coming year. We end with a glance at what's beyond video agents:Flipbook caused a minor sensation this year when it was released, but most treat it as a fun demo. Ethan takes it very seriously — with the speed and cost of inference coming down every year, the future of custom video JIT UI is closer than you think. We talked about why videogen models may become the front end of AI, how generative UI could replace traditional HTML/CSS, why world models need to be real-time, interactive, and long-horizon, and why the future of video generation may depend more on language models and agents than on diffusion alone.We discuss:* Why fast iteration mattered more than meetings* Why small training bugs can drive huge model quality gains* Why coding models may make compute the bottleneck again* How image and video models are trained with synthetic captions* The role of VAEs and latent space in frontier video models* Why image models are the foundation for video models* The tradeoff between temporal compression and real-time interactivity* Flipbook, Neural OS, and the future of generative UI* Why future interfaces may go from user intent to pixels* The hidden cost of training video models: storage, egress, and GPU hours* How step distillation and consistency models (like OpenAI sCM) makes video inference orders of magnitude faster* Grok Imagine 0.9 and large-scale audio-video generation* Why audio-video alignment is harder than text-video alignment* Ethan's definition of world models* Reference-to-video, video extension, and long-context video generation* Why xAI's research communication undersells Grok Imagine* How xAI culture shaped the speed of development* AI watermarking, SynthID, and detecting generated media* Why prompt rewriting matters for video models* Grok Imagine Agent and the rise of video agents* Why language models may unlock better video generation* Robotics, physical AI, and embodied world models* Why Ethan left xAI and shifted focus toward LLMs* Self-managed context, memory, and the next frontier for language modelsEthan He* LinkedIn: https://www.linkedin.com/in/ethanhe42* X: https://x.com/EthanHe_42Timestamps00:00:00 Introduction00:01:25 From NVIDIA Cosmos to xAI00:03:24 Building Grok Imagine from Zero to One00:10:07 How Image and Video Models Are Trained00:18:53 Video Compression, VAEs, and Real-Time Tradeoffs00:22:10 Generative UI, Flipbook, and Neural OS00:32:10 The Cost of Training Large Video Models00:37:04 Distillation, GANs, and Fast Video Inference00:41:21 Audio-Video Generation and Grok Imagine 0.900:48:34 What Makes a World Model?00:55:51 Reference Videos, Long Context, and Video Memory01:00:11 xAI Culture, Research, and First-Principles Building01:09:45 AI Safety, Watermarking, and Prompt Rewriting01:13:10 Video Agents and AI-Assisted Creation01:27:32 Why Language Models Unlock Better Video01:31:15 Robotics, Physical AI, and Embodied World Models01:32:38 Why Ethan Left xAI01:34:16 Self-Managed Context and the Future of LLMs01:38:43 Ethan's Career Path and Closing ThoughtsTranscriptIntroduction: Ethan He, Latent Space, and the Path to xAISwyx [00:00:00]: We're here in the studio with Ethan He, most recently of xAI. Welcome.Ethan [00:00:10]: Thank you. Glad being here.Swyx [00:00:11]: We're also here with Vibhu. you were first coming to us or joining the latent space world because you were working on Kosmos at NVIDIA, and you did a paper. We loved it. you presented it as well, so thank you for doing that.Ethan [00:00:23]: I've actually, I also presented the MoEs twice at latent space.Swyx [00:00:29]: How did you actually hear about us? Did we reach out to you? Is that how it worked?Ethan [00:00:33]: No, actually, I-- the community. Like I realized, oh, there is this online community that people talk about AI and also learn from each other through papers every week through the Paperclip. It's very nice.Ethan [00:00:49]: I learned a lot.Swyx [00:00:49]: I think three years stop. We haven't stopped even on Christmas and New Years. many weeks I want to stop but it keeps going.Vibhu [00:00:58]: No, that was good. I think you had posted that you worked on a paper, and I was “Oh, very cool. We have Paperclip. Present then.”Vibhu [00:01:04]: But I might have reached out to you after.Swyx [00:01:05]: you-- because it's an amateur club, right?Swyx [00:01:08]: so it's very unusual and but we have sometimes paper authors come by and actually explain the paper. Today we just did, the poolside paper, which was apparently very good.Vibhu [00:01:18]: Came out yesterday.Vibhu [00:01:19]: pretty interesting, right? Fully open. They talk about everything, systems. So it's a good one. We'll, we'll recommend people to read it.Swyx [00:01:25]: Bring us up to speed on your transition to xAI, ‘cause I actually don't even know when you joined. just like tell the, tell the story about the sort of transition.From NVIDIA Cosmos to xAI: Scaling Video and World ModelsEthan [00:01:34]: Before xAI, I was working on Kosmos world model as in-- at NVIDIA. So Kosmos is, it's a giant video foundation models that can-- that aims to simulate the world and for-- it serves as a foundation of-- for all of the roboticists to build on top of. There, once I built the Kosmos one, I realized as this thing also has a scaling law similar to language model, we need to scale up the video models further. that's, that's why I realized I need to move to somewhere with much more compute resources. That's how ISwyx [00:02:13]: Than NVIDIA?Vibhu [00:02:14]: The GPU rich came themselves.Vibhu [00:02:19]: And timeline-wise, when was Kosmo? It was pretty early, right? It was open world model, open paper, everything.Ethan [00:02:25]: It was end of twenty-four.Vibhu [00:02:28]: End of twenty-four.Ethan [00:02:30]: Then at mid twenty-five, I moved to xAI. At that time-- I joined about the time when xAI was about to build video models and in multi-model models. There were no infra, no data, and no model, and it just-- as a few engineers, we built it in three months and released the first model, Grok Imagine zero point nine.Ethan [00:02:55]: And since then, I keep working on video models and move more from training and to post-training of the video models. For example, like a reference to videos, kind of like the cameo feature and, video extensions. And, before I left, I worked on a world model, leading a small team to focus on the real-time long horizon video generation.Building Grok Imagine From Scratch in Three MonthsSwyx [00:03:24]: Can you give like a rough roadmap of okay, you're on a brand-new team. Grok previously was only text, or they partnered with BFL for their image gen stuff. What do you-- what are the building blocks, right? You have compute, data you can procure somewhere. Like just what are like the sequence of things that people should think about when you're setting up a new team?Vibhu [00:03:43]: actually even deeper, not just data you can procure. You guys had to go through getting the data too, right? So you shipped it pretty fast, but yeahSwyx [00:03:51]: three months is likeVibhu [00:03:52]: From everythingSwyx [00:03:52]: actually like very surprisingly fast.Ethan [00:03:55]: One thing I say like thanks to my experience at NVIDIA, ‘cause first time when we were building Kosmos together, we built it, for about a year. So this is like the second time I do it. Roughly have an idea, what to do. I say the most important thing is the talent. Everyone were very strong and clever, very close with each other towards a common goal. So that speed up things a lot. So you reduce the communication bandwidth among people, and everyone can work towards the same goal. It's, it's like every day there's not that much meetings on the calendar, like maybe like a, like a sync a day, and after that it's, it's just all building. It was pretty fun at that time.Ethan [00:04:47]: And another thing is that xAI has very strong foundations of like data inference, model inference, and the supporting there can help the model develop a lot. When I look at, training models, I don't so actually the top important thing is like how many, how many iterations can you do, per day? and the more iteration can you do, you can, you can train the model much faster. So if you have very strong infra and you have a lot of compute, you can, you can train these models in very short period of time. That can give you a much larger buffer to, for errors, and it also gives you the opportunity to spot more bugs.Iteration Speed, Compute, and Debugging Model PipelinesSwyx [00:05:46]: What is an iteration? Is it like a few hundred steps or what are youEthan [00:05:50]: Let's say just the train-training the model, like from acquire new data and maybe design new algorithms and train a new model, maybe at smaller scale orSwyx [00:06:01]: So cycle time for like any hyperparam that you're searching.Ethan [00:06:04]: Cycle time and tune to like eval this model. Is this model better than my previous iteration?Ethan [00:06:11]: SoSwyx [00:06:11]: So it's like before you, someone had already set this up that you can iterate very quickly.Ethan [00:06:15]: I think the foundation there is extremely good forDeveloping and research models.Ethan [00:06:23]: And often I find is it-- this is kind of boring, but like a lot of the improvements does not come from new algorithms. It comes from finding small bugs here and there in the data pipeline, in the, in the model training pipeline. Those give, those give the biggest boost to the model quality.Vibhu [00:06:46]: It's interesting, right? So you say it's like small team, less communication bandwidth, but also a lot of quality is like find little bugs. It seems counterintuitive, right? You have a lot of people, you can iron out more of those, but it's interesting to see the other side, right?Swyx [00:07:00]: I also wonder, have you-- do you try using LLMs to look for bugs? I don't know.Ethan [00:07:05]: I remember at that time it was mid two thousand and twenty-five, so it's the coding model wasn't quite there yet. I remem- I remember like December two thousand and twenty-five, it was extremely good. Yeah, I've been, I've been using it at that time. It's, it's helpful. sometimes it produce codes that are kind of difficult to maintain, even though like the first time it built something extremely fast. But it gave the, like a spaghetti code, thousands of lines that I couldn't maintain, and the LLM itself couldn't figure out what's, what's wrong and how to improve on top of it. But now I find it much better. Yeah, I want to bring up another point here is now coding models are much more efficient and can help us implement stuff much faster. Compute might become a bottleneck again because previously, like if you want to train a new model, say you want to generate new synthetic data and then or write a new algorithm, it might take a few weeks. And during that period of time, you don't-- you might not have experiments to run. But now you can build that thing within a few hours, then you can immediately train a model.Ethan [00:08:24]: Now you have to have enough compute to try all of the ideas. So compute might be the bottleneck of iterating speed again.Swyx [00:08:36]: yeah, I actually, honestly, I think it's like kind of a stressful job because you're “Well, I should be trying everything, and if I'm not, then I'm not doing my job well.”Vibhu [00:08:48]: there's also the stress of you're eating thousands of GPUs per hour, which is very expensive and, compute can go to other researchers.Swyx [00:08:56]: You got the daddy Elon toVibhu [00:08:57]: You got daddy Elon.Ethan [00:08:59]: It wasVibhu [00:09:00]: But there's still finite amount of compute, like you want to use it, you want to use it well, you want more of it.Ethan [00:09:06]: That was quite stressful indeed. Yeah, I think one thing is the-- with coding models now, like a lot of these jobs can be automated, which is much better. A second, it's a, it's a marathon, so you got to maintain good health and, a regular schedule.Vibhu [00:09:28]: It's, it's hard to hear that when you shift from zero to nothing in two months.Swyx [00:09:32]: and, I think obviously the culture at xAI is very famously, people work very hard. one thing I did want to dive into, in our-- in the notes that you, that you sent ahead of time, you had specific comments about the cost of Video Gen training. presumably this is on the Colossus-1, right? the two hundred megawatt cluster. Any whatever you want to just share on that.Vibhu [00:09:54]: I think there's, there's three things we're talking about, right? So there's Video Gen, there's also the Image Gen model that you put out. Do you want to like complete the, okay, so zero to one, you have a few months. Just what are the stages of create Image Gen model?Swyx [00:10:06]: Oh, yeah, maybe I got distracted.How Image and Video Models Are Trained: Synthetic Captions, Tokenizers, and VAEsVibhu [00:10:07]: Sorry. and then, from there's Video Gen, there's Audio Gen. Would love to get into those next. But what is that first few months like? So small team, a lot of bugs, iterations, but what does it look like? Do we take something off the shelf? Do we just get data compute? What's, what's the few months like? How do you go to state-art Image Gen model? How do you just start?Ethan [00:10:28]: I cannot comment specifically how xAI did, but it's, it's a quite standard process. I can draw some, examples from Cosmos. So mainly it's building a video model, you actually need to build a image model first. And building these two models, the data you need is a hundred percent synthetic pair of language and image or language to video. Because on the, on the internet, actually, the videos don't naturally associate with text. So you can say, oh, like on YouTube, you have the title and you have the description and the commentsSwyx [00:11:11]: TitleEthan [00:11:11]: of a video, but usually they're not relevant to the video itself. And say maybe like the video is a natural scene of mountains or something, and the title is, I'm so happy today.Ethan [00:11:26]: So they have they have no correlation at all. So the first step is to, you have to generate synthetic pair of language with the videos. So you gather videos from the internet, and you use a VLM to caption the videos. So that part, here's a question, like how do you, how do you gather VLM to begin with? So if there's noSwyx [00:11:55]: You, so you fuse the model, right? LikeEthan [00:11:57]: Say if there's no like VLM exists, like how do you generate the text to the beginning, right? It's, it's impossible.Swyx [00:12:04]: I see.Ethan [00:12:05]: In the beginning, it's like you ask human to describe the video as detailed as possible.For example, you ask them to describe everything, like all objects, all characters, and all interaction and dialogues in the, in the videos. So that's in the protocol of Cosmos labeling. We require the objective we give to the labelers was that you have to describe the video as detailed as possible, such that a blind person hears a blob of text can reconstruct what the video is like from their head.Swyx [00:12:43]: Video or image? You're talking about images.Ethan [00:12:44]: Video or image, either one of them.Vibhu [00:12:47]: This was pretty common when we went from clip and DALL-E, right?Vibhu [00:12:51]: It's all training on really detailed captioning of images. So same is applied to video, but insteadEthan [00:12:57]: same appliedVibhu [00:12:57]: of using multimodal model to pass in video images and write rich descriptions, you can alsoSwyx [00:13:04]: I think there's this traditional perspective of supervised, or, very highly human curated thing. I feel like there's a unlock with unsupervised, right? Where like you have enough to bootstrap that you can just throw common corpus on it or, whatever. like unsupervised vision and language pairing, right? Like where you just have, interspersed image and text and it just learns. To me, that is the VLM breakthrough that is different from the clip, different from the LM era.Ethan [00:13:36]: It's interesting to see that you kind of need both data.Ethan [00:13:41]: For example, for theSwyx [00:13:41]: You need it to bootstrap it up. YeahEthan [00:13:43]: for the generative model training, there's also usually like a small percentage of unlabeled data. So the model is instructed to generate a video without any text instruction. That can also help the model generalize. So after this stage of generative synthetic pair, so, one important common step is to train a compressor or a tokenizer of the image or videos. So because, if you train-- If you can technically, theoretically train image or video models on pure pixels, but the problem is that the, it's, it's a lot of tokens. So like one image, it's, a thousand by a thousand, it's like one million tokens, one million pixels. It's impossible to train transformer on that. So it's, you need to train a tokenizer, which can go from image to latent space and latent space back to image.Swyx [00:14:45]: That's why we named the podcast.Swyx [00:14:48]: But, basically, you're talking about vocabulary science.Ethan [00:14:50]: so vocab.Swyx [00:14:51]: And so, what is, what is imp-- like a million is impossible?Ethan [00:14:54]: In generative models, the vocab is continuous. It's a continuous space. We can think about like you map an image to a vector. It's a, it's a fixed length vector. It's sixteen or forty-eight, something like that. And then you map that vector back to the image space. And the mapping is, has-- The mapping is patch-based. So you say you haveEthan [00:15:22]: a sixteen by sixteen patch and you match, you map that patch of pixels into this latent space.Swyx [00:15:29]: We've covered thisVibhu [00:15:30]: This is like the vision transformersSwyx [00:15:32]: VAEs,Ethan [00:15:33]: VAEs.Vibhu [00:15:34]: You basically compress your input, you do your generation, you're reasoning all that generation in smaller dimension, and then you project back out.Swyx [00:15:43]: VAE is a form compression, but I think the for me, the patching thing is from VIT, right?Ethan [00:15:48]: You can make those.Swyx [00:15:49]: Literally the, yeah, the paper is titled like sixteen by sixteen is all you need. something like that. and then I think also, people make a lot of comparisons with this kind of patching with convolutions.Swyx [00:16:02]: Which is you're, you're kind of re- reconstructing the old paradigm with the new.Ethan [00:16:05]: Actually, in VAEs, there are, there are both convolution networks and transformers. You can actually do both.Ethan [00:16:14]: After this VAE, so what you've got is you've got latent space tokens and you've got the language tokens. So now the training of the diffusion transformer, usually generative models use diffusion transformers. It is actually quite standard. It's, it's very similar to how you train a language transformer models. It's not that much difference. It's just the tokens, the visual tokens in, visual tokens out. The only difference is there's a denoising process. So you train the model to unmask some of the noise. So you add, you add random noise to the visual tokens, and then you train the model to remove those noise to generate the clean tokens. Any inference, the model can iteratively remove noise from a hundred percent noise.Swyx [00:17:12]: And then there's also, to speed things along on the tech tree of diffusion, there's CFG, and then there's, there's also, latent diffusion that, there's, there's someone in there. I think, somewhere along the line, obviously, like stability and all these other guys, pioneered a lot of this, architecture. I don't know if you want to get into that or just, or do the video side up to you.Bootstrapping Video from Image Models and Temporal CompressionEthan [00:17:37]: After you train such model, such image model, the reason it's a, it's a foundation for video models is that image models are cheaper to train, and they have much denser connection between language and text. So, sorry, language and images. For example, you train a billion, you train on a billion images, and there's a mapping from the text to the image. And the cost to train the same, like the, a billion, a billion text to a billion videos, that's much more expensive because videosNaturally have more tokens than images. Because the diffusion models, their understanding of, language purely come from this mapping. So if you don't have enough mapping, so if you only train on like a ten million videos or something, there-- you might not see enough language tokens in your training, so your model does not understand human intention enough. So that's why you really-- you train-- you first train this image diffusion models, and then you bootstrap the video model from there.Swyx [00:18:53]: One thing I did want to ask, because I-- actually, I think you're, you're the first per-- video model person I've ever talked to, I think. we've, we've like talked to Luma and all those folks. There's all these tricks in video compression where basically frame by frame there's not that much difference, so actually you don't have to regenerate or save the whole frame, right? but I think MP4 compression or something else like that.Swyx [00:19:16]: is it tempting to use that? Or as far as I can tell, everyone just treats it as, “No, we would just generate every frame.” Is that roughly the state-art?Ethan [00:19:27]: There are a few different approaches. Let's say first, like you want to just directly use MP4 compression and use that as the tokens for the transformers to train, right? So people actually have tried that, but the main challenge is the latent space for the MP4 tokens were not, were not very comprehensible for the models. It's, it's extremely hard to train on that. And there's aEthan [00:20:01]: So that's why they created VAEs, which creates more continuous, latent space, so the models can understand that latent space and learn from it much easier. Even within the VAEs, there are different difficulties of the latent space. So you can imagine something the simplest, the most naive VAE is like you have an image, and you just shuffle all of the images into a, into a vector. So you don't need to train any VAEs, right? But that latent space is extremely hard for models to train on top of. That's why there are some debate on like how do you compress the tokens. So you mentioned like you can compress frame by frame. Also, you can compress, the temporal dimension.Ethan [00:20:52]: The difference is if you compress the temporal dimension, you get a much higher compression rate. Because there's temporal redundancy between frames, because, this frame and the last frame, likely they are mostly similar, so there's only some small difference. for example, I think in 12.1 VAE, they have like a eight by eight by four compression rate. So the four temporal tokens are compressed into one tokens. That can save a lot of, save a lot of the context length. If you do it frame by frame, you have to do maybe like eight by eight by one. Your context length will be four times larger. That being said, the benefit of the frame-- per frame compression, we might come back to this later, is, real-timeness and interactivity. ‘Cause if you, if you strain the output of the model, frame by frame, you can-- the model can respond to any user request immediately. So if you have like a temporal four compression, four times compression, thenSwyx [00:22:06]: It might be laggyEthan [00:22:07]: there's a lag there in nature.Swyx [00:22:10]: So you're very pilled on this. let's just go ahead and bring it up ‘cause we have the visual prepared anyway. There's some frontier applications of real-time video gen. So Flipbook is one of the examples that went viral recently, right? What is Flipbook?Real-Time Generative UI: Flipbook, Neural OS, and Diffusion Front EndsEthan [00:22:23]: Flipbook is kind of like a web brow- web browser. You can see like it has the web bro- browser UI on top. The difference is all of the UIs are generated by generative image model in real time, and anything here are fake. But you can, you can explore inside this wor- this imaginary world. Say like we-- here we have engineering the Great Pyramid. Like the model generates this for us to understand how it works, and if we want to navigate around and understand further, we can click on some of the, some of the description here, and the model will generate a new page, new subpage describing the details we want to know about.Swyx [00:23:14]: So it's basically kind of we're playing a video, but it's pausing for our next interaction, and then it just plays the next thing based on our interaction.Swyx [00:23:23]: Which is kind of cool.Vibhu [00:23:25]: and you kind of decide your story. So this was, how do you make a pyramid? levering technique seemed interesting, right? It shows how do you take Okay, I want to know what is thisSwyx [00:23:35]: The demo, the demo tweet had more animation between frames.Vibhu [00:23:38]: I think it's just skipping,Swyx [00:23:39]: Oh, it's just skipping a lot of frames.Ethan [00:23:40]: they also have a video modeVibhu [00:23:42]: It takes a lot. There's a lot of peopleEthan [00:23:42]: but, a lot of people are using it.Ethan [00:23:45]: So it's not available.Vibhu [00:23:46]: There's a live video stream. We can try,Swyx [00:23:50]: So this is an example of the kind of future that you see at the extreme. We don't-- we're obviously not in it today.Swyx [00:23:56]: But in a world where inference is completely free this is better than generating code and text?Ethan [00:24:02]: So this is, this is a final state of where Viva will be at for word model, I think. Imagine internet doesn't exist, and then you type in google.com. Like what should, what should, what should a model show you?the model can imagine something, and this is what the model imagine. And these web pages, they completely do not exist. So I think as the inference costs come down, we are going to have generative UI for everything. If you think about how the coding model works, so they write code for a web page, and they render the code might be con- converted into binary, and the binary render the pixels on the screen. So we in machine learning, every time we have some breakthrough, obviously it's, it's more intuit. So why don't we have like user instruction to the pixel directly? So the generative UI will be user intention to the pixels directly. And say like even if I want email, let's say everyone have the same interface, but I want, I want it slightly different. I want the email to show to me like a TikTok, so I can swipe left and right for the emails. And or maybe you want something else. We can have completely different things. Or like I have I'm looking at, Instagram stories, and I don't like the Like button. I always may click it. And, generative UI resolved it. So it's going to be a revolutionary replacement of the interface. So in the future, we might have much more powerfulEthan [00:25:50]: LLMs and coding models running behind the scene. And in the, in the front-end, the diffusion model will actually be the front-end to show stuff to you. That's how I imagine it.Swyx [00:26:02]: Diffusion front-end, deterministic back-end.Swyx [00:26:04]: Something like that. I find that very expensive, but,Vibhu [00:26:08]: I find it interesting you called LLMs writing code on the back end deterministic, but okay.Swyx [00:26:14]: you write it onceVibhu [00:26:15]: Compare it toSwyx [00:26:16]: And then you execute.Ethan [00:26:17]: If you think about the cost, say, let's say H100 costs $1 per hour, and if you use this eight hours a day and thirty days, so, every month you're paying this two forty, you'll actually not wanna pay for that. That's even more expensive than Cloud Code Max. But if you think about the compute costs come down like two times every year, and I think the future will likely arrive like within few years.Vibhu [00:26:49]: It's everything, right? compute cost comes down, compute gets faster, model gets smarterEthan [00:26:54]: More efficientVibhu [00:26:54]: model gets smaller.Swyx [00:26:55]: I don't know why you say two times, ‘cause I think it's like 100 times. In language models, it is roughly one hundred to a thousand times every twelve to eighteen months, for the same given level of LMSys, ELO.Vibhu [00:27:08]: That's a net of everything, right? That's model performance alongside compute. So different than just compute costs come down. But, a very interesting future.Swyx [00:27:19]: So the web designers will have to shout out that accessibility is an issue, right? how do you deal with screen readers or whatever. But yes, this is higher bandwidth storytelling than anything you can possibly generate with code, right? So I think that's the rough idea.Ethan [00:27:34]: And I'd like to add a little bit that so human naturally have the maximum bandwidth when we are looking at things, look at videos, and we also have maximum output bandwidth when we are talking. So in the future, it might be something like we talk to AI models, and the AI model responds back with a generative UI. So that would be the maximum input and output bandwidth to interact with AI models before neural link happens.Vibhu [00:28:06]: And it's also very custom, right? Some people are very visual, some people are not as visual, right? They prefer the text. But the best thing about generative UI, right, it can also be text.Swyx [00:28:17]: There's another project that we wanted to highlight, which is the Neural OS. Kinda similar idea, but here you're literally operating, simulating an operating system with a video model.Swyx [00:28:27]: and you can play Doom, you can do Firefox. I find this like mildly less impressive, obviously, because it's an OS that I can run.Swyx [00:28:37]: But here everything is imagined.Vibhu [00:28:40]: I was, used to the Command+W to close the Firefox tab. It didn't crash. That's why I saidSwyx [00:28:45]: It's too immersive.Vibhu [00:28:46]: It's, it's too immersive for me.Swyx [00:28:47]: Too immersive.Vibhu [00:28:48]: I wanted to close the tab.Vibhu [00:28:49]: But yes, I can play generated diffusion.Swyx [00:28:51]: this is shockingly fast.Swyx [00:28:54]: Because I remember there was a demo about like maybe one to two years ago. Someone tried to do the first-person shooter with a image model. There was no consistency. It was very slow. But here it looks like realistically it's-- this is Doom.Vibhu [00:29:07]: I think there's two sides to that, right? There's okay, what is running a game? The heavy part of it is actually the game engine, all the lighting, all that stuff, the graphics. This is just kind of video, right? Like we've solved consistency. This is still, it looks like a few years old image generation. There's some temporal consistency, but it's, it's kind of just images stitched together as frame video. But it's a good visual representation to pi- to picture the future you wanna see, right? that's, that's what I see in these more so.Ethan [00:29:38]: This reminds me of how the video models gets better and better. So Neural OS is kinda if you just look at it feels like it's just a crappy version of the, like the Windows we could have, right? And, but the difference is, so the model, this model is overfitted on the existing operating systems. It can generate nothing different than that. But it's actually also similar to video models. So when we are training these video model, image model, we train them on internet. There's no imaginary supernatural stuff on the internet. But once we train this model, you can prompt the model to generate something supernatural that have never existed in the data set. So if you train your Neural OS or neural computer on the standard screen recordings on the entire internet. The model can imagine completely new interface to interact with the computer.Swyx [00:30:43]: This is one of those things that is magical to me. usually generalizing out of distribution is bad, but somehow we have learned some kind of internal world model that you say, this plus, but it looks like rainbows and butterflies, it'll do it and it will kind of make sense.Swyx [00:31:03]: So yeah, that's kind of cool. Yeah, I don't know if there's any comment more on there. I do, I do wanted to, I did wanted to touch a little bit more on the model architecture stuff, which I think you were getting. It's, really fascinating. We don't get a chance to talk about this enough. So one of the papers that we covered, we've covered every annual, segment anything release. and I don't know if you follow-- you're a computer vision guy, so youEthan [00:31:26]: I knowSwyx [00:31:27]: . So they did memory attention, which is kind of interesting. And I always think, anything where you can, across the temporal dimension, keep some consistency, I think it's, very fascinating, and I don't know if Basically, does that-- the CV side bleeding into video gen side, I think is underexplored, right? we talk about it for labeling, but actually you can borrow the architecture itself.Ethan [00:31:50]: There's, there's also complete different approaches, right? you brought up the term world model, so we went from video model to world model. There is diffusion, but there's also other approaches that people are doing. So maybe we get into those after as well,?Swyx [00:32:03]: He has a whole definition of world models and stuff. I feel like we threw a lot at you. Whatever you want to comment on.Why Video Models Are Expensive: Storage, I/O, and Training ScaleEthan [00:32:10]: I think one thing that we should actually comment back on is okay, so we were talking about the steps to train image gen to video model. One thing we don't see as much of is okay, you brought up the delta in training data, right? SoEthan [00:32:24]: you won't have as much a video model might not generalize, but what is the cost of training a large video model? So we know for LLMs roughly, okay, even like the poolside thing that came out today, right? It's a Gemma level model trained on roughly forty trillion tokens at this many H200s over this much time, right? You can see what is the exact cost of that. So how many GPU hours over how much H200 costs? So how do we do the back-end math of, same thing for video models, image models. How do you, how do you kind of break that down? I can share some back-envelope calculation. So surprisingly, video models is-- the cost is very-- is comparable to language models and obviously the largest scale is language model, maybe like a medium scale to language models. I said just storing the videos alone, it costs a lot. You can, you can maybe look up on AWS or something.Ethan [00:33:20]: You really, say if you have a billion videos and let's say, let's just say like each video, like five megabyte, then you need five petabyte to just store those videos. And also remember we talk about you use a VAE to compress the videos, and you also need to store, typically you need to store those continuous feature, in-- also in your storage. That's also comparable size with the videos themselves. So just storing these videos and the features is tens of petabytes alone. And,Swyx [00:33:58]: I just, I just looked up the calculation. Five petabytes on S3 Standard is one hundred K per month.Ethan [00:34:05]: AndSwyx [00:34:05]: It's comparableEthan [00:34:05]: and you needSwyx [00:34:06]: AndEthan [00:34:06]: And then like tens of petabytes, two hundred K. And even more expensive is you have the ingress and egress.Swyx [00:34:13]: Oh, yeah.Ethan [00:34:14]: Like you-- through the internet. You have to just to download those videos, I believe it's, it's more expensive on AWS than just storing those videos.Swyx [00:34:25]: Storing, yeah.Ethan [00:34:25]: And each training runs, you probably need to pull them once. If you train multiple times, it's, it's even more than that. So it's like just storing the network, those costs is just, it would be a few, a few millions per month to just storing everything, not to mention the GPU cost.Ethan [00:34:45]: AndSwyx [00:34:45]: my side tangent, the compute rental, like GPU rental is very efficient. There's one side, okay, you can be XAI and build your data center. Should we not just build our, storage compute as well? LikeEthan [00:34:57]: Of courseSwyx [00:34:57]: cloud cost compared to just,Ethan [00:34:59]: You save so muchSwyx [00:35:00]: store. Yeah, exactly.Swyx [00:35:01]: Especially with like egress and stuff. So.Ethan [00:35:04]: That's a good idea, but it also comes to-- there are some of its own challenges.Swyx [00:35:09]: Of course, of course.Ethan [00:35:10]: like people who build the GPU data centers, they might not expect this much, storage. And yeah, people build storage, typically they just build it somewhere with just CPUs.Swyx [00:35:23]: I just looked it up. Five-- AWS only charges for egress, not ingress. Tier five for five petabytes is two hundred and thirty K.Ethan [00:35:32]: Even more expensive than the storage.Swyx [00:35:34]: But storing is per month, right? You check in, then you cannot check out. so it's so cool. It's okay. So there's that side.Ethan [00:35:41]: So the TLDR, my backhand mathSwyx [00:35:42]: Data is larger than you think. Yes.Ethan [00:35:44]: my backhand math of GPU hours times GPU cost is also very much, I'm missing some storage.Swyx [00:35:49]: You're also-- you're basically like also more IO bound than normal training.Swyx [00:35:55]: Yes. ‘Cause like data loading, so caching everything, it becomes super important.Ethan [00:36:00]: So in Cosmos, we did a lot of optimizations to make it not IO bound. So, speaking of the training, actually training the model, the GPU cost, if you look up like the open source model, how big these video models are, I think like LTX has nineteen B parameters. That's a dense model. And people are also exploring, MoEs, so it might be twenty B active and, like a hun- hundreds B, total. So that's, that's even-- that's similar size as medium-sized LLM models. And if you, if you look at number of tokens-Uh, we disclose that in Cosmos. It's also like tens of trillions of tokens on the visual tokens. So putting this together, the cost of, training these video models, it's actually comparable with LLMs. Not to mention, the infra is slightly different from LLM, so it might be less efficient to train these models.Inference Speedups: Step Distillation, Consistency Models, and GANsSwyx [00:37:04]: Do you get the benefits of traditional diffusion speed-up? So for, images, there's LCM, LoRAs for, fine-tuning. There's, there's a lot of stuff that's beenEthan [00:37:15]: Flow matching.Swyx [00:37:16]: there's flow matching. There's a lot of stuff that's been done. there's some overlap that applies to diffusion on the inference side and stuff or?Ethan [00:37:23]: so the difference-- the inference side is a completely different story.Ethan [00:37:28]: I think for the training side, it might be a little bit hard to reduce that cost. And for the inference side, the biggest gain is from the distillation of these models. You can-- It's called step distillation, slightly different from knowledge distillation in LLMs. So you-- Typically, for flow matching models, you need like 100 steps or something. Like a distortion model even need even more, like 1,000 steps to generate a good image or video. A step distillation is try to learn to generate fewer step from the model itself. It's kind of like now we-- you use the full model to generate in 100 steps, and then you take a model that only generate 10 steps and let that model to learn from the perfect one.Ethan [00:38:25]: why this workSwyx [00:38:27]: Strong to weak seemingly.Ethan [00:38:28]: It is. It's kind ofSwyx [00:38:29]: DistillationEthan [00:38:29]: kind of like strong to weak. the-- from the modeling perspective, the strong model, the teacher model is trying to model the image and videos of inter-internet, and that distribution is extremely complex. But the step distilled model is just trying to learn from the teacher. The teacher is a model, and the size is fixed, as the distribution is much simpler than the whole internet. That's the intuition I have why step distillation can work. So usually these models serve in productions, they only run in a few steps. In Cosmos, I believe we have, we have like four step and eight steps. If you do some simpler task, image-image translation, it can even run in fewer step, like one step in Cosmos Transfer.Swyx [00:39:22]: I think this is the same intuition that guides a lot of the consistency model work. I sent you a link for, SCM. I don't know if you covered that. To me, that was actually one of, the most impressive papers I've ever seen from OpenAI.Swyx [00:39:34]: That this is the unifying grand concept of consistency models. I don't know if you have any comments on this.Ethan [00:39:41]: So there are, there are a few different approaches,Swyx [00:39:46]: Oh, yeah. Here it is.Swyx [00:39:47]: Two steps versus twenty or 100 steps, whatever. It's already done.Ethan [00:39:52]: So there are, there are a few different approaches, for example, consistency model, and there are also Actually, we shouldn't forget GAN. So GAN, actually, that was, that was the OG ofSwyx [00:40:05]: OGEthan [00:40:05]: step distillation ‘cause it trained just one step to begin with. So actually, a lot of, uh-- For example, there's a distribution matching distillation which use, which uses GAN, as one of the laws for distillation. It-- GAN just tells you, “Hey, generate an image,” and thenEthan [00:40:31]: it has a discriminator to tell, is this image real or not? So the model, the model just need to learn one of the distribution, not the full distribution. Because in training, the model is asked to reconstruct the ground truth image from the internet, which is extremely hard. And in-- When you're training GAN, it's a step process. It's just a, “Hey, you generate image. Does this image look as real as the image from the internet?” Which is a much simpler task. And, yeah, combining a lot of these approaches together, people typically do that, like consistency model and distribution matching and GAN, and we can get these few step models.Audio-Video Generation and Time AlignmentSwyx [00:41:21]: Then there's one step I wanted to add, which is audio and video.Ethan [00:41:26]: So, Grok Imagine zero point nine, I believe it's, it's a first audio video transmodel deployed at a large scale. SoSwyx [00:41:39]: And that was your first model?Ethan [00:41:40]: that was, Grok Imagine's first model. It's, it's audio video, joint generation. I think the hard part is, the modality alignment, ‘cause before this transmodel, we have, we have text to video alignment. We have this, correspondence between text and video. Typically, most of the VLMs, they understand images and videos. Video's very rare, and they don't understand audio mostly. And if you look at the audio generation on the LLM side, you can talk to them perfectly fine, but if you ask them to sing a song or something, it typically is not very good. Also, they don't have, they don't have music either. The hard part is thatUh, actually audio has two component. It has like a discrete component, a continuous component. The discrete component is like the language.Ethan [00:42:44]: So when we speak, it's just, someSwyx [00:42:47]: It's an ASR issue, yeah.Ethan [00:42:49]: It's, it's text token with some characteristics, I would say.Ethan [00:42:54]: But musicSwyx [00:42:56]: I think the speech guys would disagree with this.Swyx [00:42:57]: Like disfluencies and then,Vibhu [00:43:00]: There's tones you can get angry.Ethan [00:43:01]: Well, I say largely.Ethan [00:43:03]: the mu- but the music is completely different. It's, it's very continuous, and you cannot model them like discrete tokens in language models. this is like the hard part for models is, not to mention we have to align text, video, and audio together.Ethan [00:43:26]: SoVibhu [00:43:26]: How?Ethan [00:43:28]: So significant-- some significant challenges are like-- So first, like we talk about as the VLMs, they cannot understand most of them cannot understand audio.Ethan [00:43:39]: So you have to have some way to do the synthetic data generation for audio. You have to caption the model, and that involve, that involve synthetic data and human data effort a lot. And not just surprisingly, most of the LLMs are very bad at recognizing, like the beat, tone, and the details of the of music. They can, they can give some general prediction of which song is this, but it's very hard to describe the details of the music. like we mentioned in image generation, like you have to describe image as detailed as possible so that someone blind can reconstruct that. So here is like someoneVibhu [00:44:32]: DeafEthan [00:44:32]: someone deaf can reconstruct how the music sounds like without actually listening to it. Maybe you can think of it need to have the-- or they call the script.Vibhu [00:44:49]: Subtitles, yeah.Ethan [00:44:49]: You gotta have all the details of the music, and the dialogue.Vibhu [00:44:55]: So is the challenge there typically stuff like music and audio, or is it just Like is there a baseline? Okay, there's enough data where we can understand, narration, conversation, but there's nuances in audio that's where you hit all the data issues or is it just from stage zero, you just do it all right?Ethan [00:45:15]: So one important thing is like the alignment. So the model, the model has to know like the video and audio, the, uh-- it has to have a time-based alignment, like at which time step the video and the audio token correspond to each other. But we actually don't have this kind of alignment for most of the other modalities. If you think about like text and image, text and video, they are loosely aligned. So you can, you can have a description of what's going on in the video, but you don't have to exactly, You typically don't have exact description, oh, at, time step one second like what happened?Vibhu [00:46:02]: It's veryEthan [00:46:03]: At time step two second what happenedVibhu [00:46:03]: coarse. Yeah.Swyx [00:46:05]: So what was the ideal time step? You have to oblate it, and then it's like four seconds or something.Ethan [00:46:09]: So that comes down to how you design the model to, for the model to be aware of as a time, as a time modality. So the model is like a time aware. And that's something pretty unique if you think about LLMs. So if you ask LLM to complete a task, say they, uh-- you ask them and they will say, “Oh, this task will probably take twelve hours to complete,” and they come back in one hour. Say “I've already spent two days on this and I've exhausted everything.”Ethan [00:46:47]: So the LLMs them-themselves, they don't have a sense of time there.Vibhu [00:46:53]: I actually don't think that's just them not having a sense of time. I think it's somewhat based, right?Vibhu [00:46:58]: Like you tell someone, “Okay, go work on this feature. Go implement this,” there's a general understanding you would have of how long that would take without LLMs working at LLM speed, right? So you think back like two years ago, if I tell you to like build me like a new front end for latent space, have a search bar, have all this, you'll estimate that it'll take a few days, right?Vibhu [00:47:19]: So you tell an LLM, “Go build this.” It'll take me a few days. But I think it's somewhat grounded as opposed to them not having the best-- Not saying that they have a great understanding, but I think that example is like you can see where it comes from, right? You're trained on all over the text.Swyx [00:47:35]: They're, they're trying to estimate what a human would say.Vibhu [00:47:37]: because that's what the, that's what the data kind of represents. It's not themEthan [00:47:41]: It came from the corpus on the internet. People have a estimate of how much time.Vibhu [00:47:45]: And not even just in direct like training samples, right? Just your world understanding of tokens of how long stuff takes, right? Go read a book. It'll take you a while, right?Vibhu [00:47:56]: Even if you do nothing but read a book, it takes a few days. So yeah, LLM, I read it took me a few hours.Vibhu [00:48:01]: It'll take me a few hours to go through this research. But this is a tangent.Swyx [00:48:05]: Somewhat, yeah.Swyx [00:48:06]: This is a train of thought I haven't really expressed until now is, which is basically like a full world model must also be recursive, meaning that the participant in the world model must also be aware that they have a world model. which is like this whole recursive thing down the, down the line. but yes, and that the world model can be wrong and that they need to update it and blah. Yeah. We've, argued this on the, newsletter as well, that there needs to be sort of recursive or adversarial world models.World Models: Real-Time, Long-Horizon, Interactive VideoVibhu [00:48:34]: just, to ask, how do you define world model?Swyx [00:48:38]: Oh, yeah, let's go there.Ethan [00:48:40]: SoVibhu [00:48:40]: So just for context, we talked about, video generation, and then there's a-- if you say there's a distinction between world models, what's your, what's your definition? How do you see the two?Ethan [00:48:53]: So disclaimer, I'm not going to debate, what is world model. Yeah. there are many definitions, so I'll just talk about my definition. Since I came from the multi-model, multi-model domain, so mainly talking from video. So world model is like real-time interactive long horizon videos. So there are three parts. so we-- let's talk about them one by one. So the so interaction, so we just, we just look at Facebook and neural computer. So the interaction part of it, so you, world model can allow you to interact with them through keyboard, mouse, and maybe also voice. So these all is-- all is a modality. You can, you can interact with the model, and the model should respond reasonably. Second part is real time. So once you, once, say, you move your mouse, if, say, the world model generate a game, how fast can the game respond? So if you're like professional CS: GO players- -my say, oh, you have to respond- He's beginner within sub ten milliseconds or- Yeah even less. So that's not most of the- No, sixty FPS. Let's go. Oh, three hundred FPS. Oh, five hundred FPS. Wait. okay, yeah. I didn't do the math, but yeah, okay. Uh- Yeah, three hundred FPS, that's a three millisecond. So you have to respond- Oh, s**t. Okay. YeahEthan [00:50:29]: within a millisecond. Most of the video models cannot do that. Yeah. And, but if you, say, if you have a video model that is, say, like a digital human, the response time might be more generous. Maybe typically, for real-time voice interaction, it's like two hundred millisecond. So that's, that's much more generous. But even two hundred millisecond is pretty, it is pretty tricky, ‘cause remember we mentionedEthan [00:51:01]: you have this, temporal compression coming from the VAE. So if you, if you don't compress the temporal dimension, your sequence length is going to explode. So if you want to have this real-time, real-timeness in your model, you have to do is one context problem. And the third part is long horizon, ‘cause we-- if you're not going to just play with, video games just, a few seconds, most video models only a few seconds. We're going to play with minutes, hours. The model have to be able to generate long-form content.Ethan [00:51:42]: So putting these three together, it's, real-time, long horizon interactive videos. I think the final state will be, for example, like a video, a video version of Playbook, where you can, you can interact with, a neural computer. You move your mouse, and you click on the generative interface, and it will reply to you through pixels- generating in real time. But getting there, it's, it's a very long way to get there. So one of the first step, at Grok Imagine, where I led a small world model team there, was to build video extension. So, video extension- it's the first step of interactivity. Yeah. It's, it's the first step. Yeah. So it's the first step- You have it here, video editing, yeah. Yeah. Yeah. So the first step is because, this unlocks long horizon videos. Typically, for most of the video generation models, you give it a prompt or an image as an initial frame. You generate video, that's it. That's just, one time, done. And some creators would try to, use the last frame as a first frame for the second video. It can-- sometimes it works, but if you do it a few times, it says the quality would decrease. And- It doesn't have that context- Yeah over the full video, so the temporal- Yeah, exactly. Yeah, ‘cause you only gave it the last frame, of course, right? Yeah. Exactly. And- it's actually a pretty fun hack. if you've seen like- Oh, no, he's saying something better. Yeah. And for example, like Vue, I remember Vue 3 has like a second context of the last video. It is slightly better than using the last frame, but it has the same problem-- similar problem that it, the quality would decrease. if you extend a few times to, one minute, the video quality would look much worse than the first video. Second, another problem is that the model doesn't have long-range knowledge of, what's happening before. Say, if they generate some dialogue, some, two people speaking, and their voice might change, over some time, especially if the second conditioning, it does not cover the previous context. So these are the core challenges. So the Grok Imagine video extension, it has historical context of all of the previous generated videos. It can, It has, it has the context of, who is speaking and what objects have appeared and everything, having that to generate the next video. So if we naively do this, you can imagine, just, put all of the previous history video tokens into the context. The context lens will easily explode. Especially for video models, that can be like a few, a few million context, I would imagine- context lens. Yes.Yeah.Swyx [00:54:58]: Let's run with that.Ethan [00:54:59]: for example, like in Cosmos, I think just five seconds of video is like a fifty K or sixty K number of tokens. So like if you do, if you do fifty second, that's a five hundred K tokens. If you do longer than that, easily explode. This long horizon, problem was the first step we're trying to solve world model. It turns out people, yeah, people love video extension. Like a lot, a lot of the creators love using video extension to create longer form videos. This is the part I liked that you have a, you have an intermediate step toward the final goal instead of just a straight shot to the final version very much.Swyx [00:55:48]: But I can see you have a strong vision of where we want to end up.Long Context, Redundancy, and Efficient Interactive VideoVibhu [00:55:51]: Does it seem like it's an efficiency issue? okay, we're at a few million tokens context,. If you draw the parallel to language models, we had very short context, two thousand, eight thousand, then, you scale it up one million, ten million. sure, there's effective context, but at the end of the day, it's just what's it worth? sure, there's a whole training data side. In video, it might be slightly easier ‘cause we have a hundred million token video, right? Just take a movie with the full context there. Like is this efficiency from an inference standpoint that like it's expensive, but we know how to solve it? Or like why is this not the approach? So like my broader point was on your second point of world models, you say it needs to be interactive and live, right? You should be able to play a game and see the interaction live. So one thing I see with research is a lot of what you actually serve is different than what you build, right? So we talked about distillation. You train big model, you distill it, you do quantization, speculative decoding. We do all this stuff to serve it efficiently. Should we not just have a solution, like a world model that can interact well, do inference optimization, serve it, distill it secondary, so make it real time after you solve it? So like a-- another parallel is say, continual learning, right? What we need is someone to solve it and show it works inefficiently. Give it a few years, people will make it efficient. Same thing with regular attention, right? It worked. Over a few years, people have different forms of attention, and we've scaled it to be efficient at log context,? So kind of two things there, right? One is it seems like it works. You've scaled it. Can we not just scale it a lot more efficiently over time? Do we need a separate approach if this works? And same thing with interaction, right? if we can get it done, like if we can solve some way that it works, we can solve making it more efficient from an inference standpoint later.Ethan [00:57:53]: that's actually a very good point. So in videos, there's actually a lot of redundancies. So we solve a lot of the pixel redundancy from VE, but there's more redundancy in long range and long horizon videos. Say, if a character appear in the first clip and then it disappeared, it only reappear at the end of the video, you probably don't need the-- the context, like in the middle of the generation. So you only need that character, where you need. So that's why, I helped build another feature. It's a reference video.Vibhu [00:58:36]: Is it here?Swyx [00:58:36]: is it the same model release or different one?Ethan [00:58:39]: It's a different one.Ethan [00:58:41]: You probably need to search onSwyx [00:58:43]: I'll find itEthan [00:58:43]: X reference to video.Ethan [00:58:46]: So reference video allow you to like upload up to seven images as condition and generate the video. Say, if like I want-- it can, it can be characters or objects or even scenes. Say like I want, I want condition on, Sean's selfie and holding a bladeSwyx [00:59:07]: We have a dogEthan [00:59:08]: or whatever.Swyx [00:59:08]: We put the dog in the thing.Ethan [00:59:09]: you can put them there and the video models will generate the video from and copies the context over. So that can solve a lot of the problems there, like the long context problem. It doesn't need to have a very long context, but it's-- I feel like it's an intermediate solution. The modelSwyx [00:59:29]: It's cheating.Ethan [00:59:30]: the model should be able to like selectively know, where should I draw the references. So say if I want to generate a movie, I generate it autoregressive, like a ten second at a time or something. And now this character appear, I can look back to where it first appear and, bring that back. Yeah, this one, I put the references. Yeah, that's, Optimus, Einstein myself, Annie.Vibhu [01:00:02]: Oddly enough, I used Grok Search to find it, and it pulled your LinkedIn post. But yeah we found it.Ethan [01:00:08]: Interesting.Vibhu [01:00:10]: ButxAI's Underrated Work, Culture, and WatermarkingSwyx [01:00:11]: this is a problem. This is not your fault, but like XAI doesn't communicate all this work that you do very well because they just have the model release and then that's it. But actually, these details are very good.Swyx [01:00:22]: As far as I understand, everything you just described is state-art, like no one else has done it.Vibhu [01:00:30]: A lot of-- yeah, I have a lot moreSwyx [01:00:32]: And then, and then you just put this blog post with the cookies. I'm this is not enough,?Swyx [01:00:37]: but I, obviously this is like the high level numbers that people want to know. But no, okay, soVibhu [01:00:42]: And I wonder, like part of that is also some labs don't share research into what happens. And ifSwyx [01:00:50]: No, but this is literally bragging about how good they are, right?Swyx [01:00:54]: Like, why would you not say that you are capable of extending with full context? this is not a secret sauce. This is like we did the work. yeah, I don't know.Ethan [01:01:02]: different labs have slightly different communication styles.Swyx [01:01:07]: Anyway, if anyone from XAI is listening we are always happy to help you tell your story. Yeah, okay, so you did references, and I think, I think kind of the point you're, you're making is it is sort of like a kludge, right? this is-- you can do seven, but what about 100?Swyx [01:01:23]: Right? Then you need a completely different thing.Ethan [01:01:26]: So I think it's-- this is, a mechanism to, select the context from the history, and you might not put the entire history into the context. for example, there's a paper called Frame Pack, which haveEthan [01:01:41]: a heuristic that the latest history, the last one second, I put the entire history, and the history before that, I would, compress it and makes the video smaller. So they follow this pattern, this build overall pattern that the maximum sequence length is fixed. So the further you are from the current frame, you have a smaller image. So this is just a heuristic. I think it can be more automatic. The model is aware like which history part of it can be select. So this part of the research is actually being actively, worked on by a lot of people. It's also quite interesting. I feel this is actually, this part of long context is a little bit ahead of the LLM part.Ethan [01:02:31]: So for example, like in LLMs, if you-- so contexts keep growing. Let's say if you call tool and the tool call history is extremely long, that's still in context, and keep growing, keep growing. Even if you switch the topic to something else, the whole context was there. There are some agentic harnesses that help you to, say, prune the tool results and, prune Like when you, when you query a file, only show like the top 200 lines or something. Those were very heuristic-driven.Swyx [01:03:08]: For listeners, we did a write-up on the cloud code, leak where there are eight different kinds of pruning, including like you prune the tool results and all that. So you can, you can read up on that kind of thing.Ethan [01:03:17]: I think, one breakthrough in continual learning might be like a way to automatically, manage its own context.Swyx [01:03:27]: These are all heuristics, and they will be replaced by machine learning.Ethan [01:03:30]: InterestinglyVibhu [01:03:32]: TheEthan [01:03:32]: the same thing is being researched in both LLMs and video models.Vibhu [01:03:36]: The interesting thing is also like in the paper you showed, it's actually happening at the model level, right? Compared to like language models, sure, we have base attention, but we'll do our own compression, we'll do our own pruning, which is separate from model error.Vibhu [01:03:49]: Eventually, it all just boils in, hopefully.Swyx [01:03:52]: I think this is a form of like attention, but like also know sort of reasoning attention. I feel like that's different than normal attention.Swyx [01:04:03]: Does that, does that make sense?Ethan [01:04:04]: It's, it's different in the sense that attention, not to mention, set sparse attention aside,

  EU AI Act 2026: Was jetzt gilt und was noch auf dich zukommt   Der EU AI Act ist verabschiedet — aber er wird bereits überarbeitet, bevor er vollständig in Kraft ist. Was das konkret für dich als Selbständige oder Unternehmer bedeutet, welche Regeln schon heute gelten und worauf du dich bis 2027 vorbereiten musst, erklärt Rechtsexperte Philipp Hacker im Koerting-Institute-Podcast. Hier die wichtigsten Punkte im Überblick.   Philipp Hacker auf LinkedIn: LinkedIn - https://www.linkedin.com/in/philipp-hacker-078940257/   Was bereits gilt: Verbote, KI-Kompetenz und GPAI-Regeln Einige Teile des AI Act sind schon jetzt verbindlich — darunter das Verbot von Emotionserken-nung am Arbeitsplatz und die Pflicht, dass alle, die mit KI arbeiten, ein Grundverständnis über tech-nische und rechtliche Zusammenhänge mitbringen. Wer ein bestehendes KI-Modell unter eigenem Namen vermarktet, kann außerdem rechtlich als Anbieter eingestuft werden. Neue Zeitpläne: Chatbots, Wasserzeichen und Hochrisikopflichten Ab dem 2. August 2026 muss jeder Chatbot zu Beginn einer Interaktion klarstellen, dass es sich um KI handelt… Nutzer dürfen nicht den Eindruck bekommen, mit einem Menschen zu sprechen. Die Regeln zu Watermarking und Labeling folgen erst ab Dezember 2026, die Hochrisikopflichten zu Risikomanagement, Datengovernance und menschlicher Aufsicht sogar erst 2027 oder 2028. Diese Verschiebungen sind das Ergebnis der laufenden AI-Omnibus-Überarbeitung, mit der die EU den AI Act bereits nachschärft, bevor er vollständig in Kraft ist. Deepfakes und Recruiting: Die zwei Themen, bei denen du jetzt handeln solltest Deepfakes sind weiter gefasst als oft angenommen: Darunter fällt nicht nur das gefälschte Perso-nenvideo, sondern auch ein KI-generiertes Produktbild auf deiner Website oder substanziell bear-beitete Marketingfotos… sobald KI wesentlich eingreift, gilt Kennzeichnungspflicht. Draft Guide-lines zu Deepfakes sind seit dem 8. Mai 2026 im Entwurf verfügbar und erklären anhand von Bei-spielen, was kennzeichnungspflichtig ist und was nicht. Fazit Der EU AI Act ist kein abstraktes Brüsseler Projekt mehr… er ist bereits in Teilen geltendes Recht und betrifft auch Selbständige und kleinere Unternehmen direkt. Die unmittelbar relevanten Punkte sind überschaubar: Chatbots kennzeichnen, keine Emotionserkennung gegenüber Mitarbeitenden, KI-Kompetenz aufbauen und KI-generierte Inhalte als solche ausweisen. Wer diese Entwicklungen regelmäßig verfolgt, ist nicht nur gesetzeskonform, sondern baut echtes Vertrauen bei Kunden und Partnern auf.     Noch mehr von den Koertings ...  Das KI-Café ... jede Woche Mittwoch (>350 Teilnehmer) von 08:30 bis 10:00 Uhr ... online via Zoom .. kostenlos und nicht umsonstJede Woche Mittwoch um 08:30 Uhr öffnet das KI-Café seine Online-Pforten ... wir lösen KI-Anwendungsfälle live auf der Bühne ... moderieren Expertenpanel zu speziellen Themen (bspw. KI im Recruiting ... KI in der Qualitätssicherung ... KI im Projektmanagement ... und vieles mehr) ... ordnen die neuen Entwicklungen in der KI-Welt ein und geben einen Ausblick ... und laden Experten ein für spezielle Themen ... und gehen auch mal in die Tiefe und durchdringen bestimmte Bereiche ganz konkret ... alles für dein Weiterkommen. Melde dich kostenfrei an ... www.koerting-institute.com/ki-cafe/   Mit jedem Prompt ein WOW! ... für Selbstständige und Unternehmer Ein klarer Leitfaden für Unternehmer, Selbstständige und Entscheider, die Künstliche Intelligenz nicht nur verstehen, sondern wirksam einsetzen wollen. Dieses Buch zeigt dir, wie du relevante KI-Anwendungsfälle erkennst und die KI als echten Sparringspartner nutzt, um diese Realität werden zu lassen. Praxisnah, mit echten Beispielen und vollständig umsetzungsorientiert. Das Buch ist ein Geschenk, nur Versandkosten von 9,95 € fallen an. Perfekt für Anfänger und Fortgeschrittene, die mit KI ihr Potenzial ausschöpfen möchten. Das Buch in deinen Briefkasten ... https://koerting-institute.com/shop/buch-mit-jedem-prompt-ein-wow/   Die KI-Lounge ... unsere Community für den Einstieg in die KI (>2800 Mitglieder) Die KI-Lounge ist eine Community für alle, die mehr über generative KI erfahren und anwenden möchten. Mitglieder erhalten exklusive monatliche KI-Updates, Experten-Interviews, Vorträge des KI-Speaker-Slams, KI-Café-Aufzeichnungen und einen 3-stündigen ChatGPT-Kurs. Tausche dich mit über 2800 KI-Enthusiasten aus, stelle Fragen und starte durch. Initiiert von Torsten & Birgit Koerting, bietet die KI-Lounge Orientierung und Inspiration für den Einstieg in die KI-Revolution. Hier findet der Austausch statt ... www.koerting-institute.com/ki-lounge/   Starte mit uns in die 1:1 Zusammenarbeit Wenn du direkt mit uns arbeiten und KI in deinem Business integrieren möchtest, buche dir einen Termin für ein persönliches Gespräch. Gemeinsam finden wir Antworten auf deine Fragen und finden heraus, wie wir dich unterstützen können. Klicke hier, um einen Termin zu buchen und deine Fragen zu klären. Buche dir jetzt deinen Termin mit uns ... www.koerting-institute.com/termin/   Weitere Impulse im Netflix Stil ... Wenn du auf der Suche nach weiteren spannenden Impulsen für deine Selbstständigkeit bist, dann gehe jetzt auf unsere Impulseseite und lass die zahlreichen spannenden Impulse auf dich wirken. Inspiration pur ... www.koerting-institute.com/impulse/   Die Koertings auf die Ohren ... Wenn dir diese Podcastfolge gefallen hat, dann höre dir jetzt noch weitere informative und spannende Folgen an ... über 500 Folgen findest du hier ... www.koerting-institute.com/podcast/   Wir freuen uns darauf, dich auf deinem Weg zu begleiten!

Con lo sviluppo esponenziale dell'Intelligenza Artificiale Generativa, è diventato sempre più difficile distinguere i contenuti creati dalle macchine da quelli prodotti dagli esseri umani. Se 5 o 6 anni fa eravamo già preoccupati dal dilagare di fake news e video deepfake, oggi la situazione è ancora più grave. I social sono invasi da “AI slop”, mentre il web si riempie di applicazioni sviluppate interamente con l'IA, spesso prive di supervisione adeguata. È quindi diventato fondamentale sviluppare sistemi e tecnologie in grado di permettere a utenti e applicazioni di riconoscere se un contenuto – sia esso un video, un testo, un'immagine o una musica – è stato generato totalmente o parzialmente da una macchina. Ma è ancora possibile farlo? E quanto sono affidabili i sistemi che stanno nascendo per questo scopo? In questa puntata proviamo a rispondere a queste domande.Nella sezione delle notizie parliamo di una nuova tecnica CRISPR contro i tumori, della class action contro Apple per le promesse non mantenute su Siri e Apple Intelligence e infine dell'avvio della produzione del camion elettrico di Tesla.--Indice--00:00 - Introduzione01:12 - Una nuova tecnica CRISPR contro i tumori (ANSA.it, Luca Martinelli)02:58 - Siri nel mirino di una class action (Wired.com, Davide Fasoli)04:10 - Tesla avvia la produzione del tir Semi (DMove.it, Matteo Gallo)05:37 - Riconoscere contenuti generati dall'IA. È ancora possibile? (Luca Martinelli)18:35 - Conclusione--Testo--Leggi la trascrizione: https://www.dentrolatecnologia.it/S8E19#testo--Contatti--• www.dentrolatecnologia.it• Instagram (@dentrolatecnologia)• Telegram (@dentrolatecnologia)• YouTube (@dentrolatecnologia)• redazione@dentrolatecnologia.it--Brani--• Ecstasy by Rabbit Theft• Falling For You by SouMix & Bromar

Stegawave, an Irish technology company specialising in forensic watermarking for video content, has announced the launch of its anti-piracy platform for live sports streaming. Using a proprietary watermarking algorithm to embed invisible patterns into live streams, Stegawave identifies piracy sources in real time, shutting down illegal redistribution within minutes of detection. The platform integrates with existing streaming workflows and distribution systems, requiring no changes to a content owner's existing infrastructure. Stegawave customers also have the option to push alternative content or messaging to the illegal stream destination. Live sports piracy costs the global broadcasting industry billions annually, with the U.S. Chamber of Commerce estimating the impact on the US economy alone at more than $29 billion per year. Illegal IPTV services, often referred to as 'dodgy boxes', restream content within minutes of broadcast, and legacy content protection technology cannot identify the source of a leak. For smaller and mid-tier organisations, the financial impact is proportionally even greater, as lost subscribers directly threaten the viability of grassroots and regional sports attendance and coverage. The announcement follows a successful deployment with Clubber TV, a leading live sports platform. Sports broadcasters worldwide are facing persistent piracy from illegal IPTV services that directly impact subscription revenue and the financial viability of the coverage itself. Stegawave was deployed across live broadcasts, detecting pirated streams and identifying the specific subscriber accounts responsible. Stegawave achieved a 100% detection rate across all streams, and because many illegal IPTV services share the same source account, blocking a single compromised account simultaneously disabled multiple pirate streams, amplifying the impact of each enforcement action. "Piracy is a massive threat to the sustainability of sports broadcasting at all levels, including grassroots coverage. This technology is potentially game-changing for us to ensure, following significant rights fee investments, that fans are only watching Clubber games on our platform," said Jimmy Doyle, CEO, Clubber. "It's been a pleasure to work alongside Clubber in stopping piracy of their premium matches. The work we have done together has helped us improve the Stegawave product and also resulted in new features. With the increase in illegal streaming not just in Ireland but worldwide, there is real momentum behind tackling this problem, and Stegawave can play a key role in tackling it both here and internationally. We look forward to supporting sports rights holders and broadcasters across the globe in recovering lost revenue and protecting their premium content." said Sean Fahey, CEO, Stegawave. Stegawave is now available for streaming platforms, sports broadcasters and rights holders who are looking to protect their premium content and maximise their Pay-Per-View and Subscription revenues. The Stegawave team will be attending NAB Show in Las Vegas from 18-22 April 2026, showcasing the platform to technology partners, content owners and wider media. See more stories here. More about Irish Tech News Irish Tech News are Ireland's No. 1 Online Tech Publication and often Ireland's No.1 Tech Podcast too. You can find hundreds of fantastic previous episodes and subscribe using whatever platform you like via our Anchor.fm page here: https://anchor.fm/irish-tech-news If you'd like to be featured in an upcoming Podcast email us at Simon@IrishTechNews.ie now to discuss. Irish Tech News have a range of services available to help promote your business. Why not drop us a line at Info@IrishTechNews.ie now to find out more about how we can help you reach our audience. You can also find and follow us on Twitter, LinkedIn, Facebook, Instagram, TikTok and Snapchat.

Is Generative AI moving too fast? From viral deepfake videos to powerful coding assistants, AI is reshaping our world at a breathtaking pace. But with this power comes immense risk: to our privacy, to intellectual property, and even to our ability to tell what's real. How do we navigate this complex new landscape responsibly?In this episode, Allen sits down with Maggie Engler and Numa Dhamani, authors of "Intro to Gen AI, Second Edition" and veterans in the fields of cybersecurity and trust & safety. They pull back the curtain on how these powerful models are built, the societal impact they're having, and the urgent conversations we need to have about data governance, AI agents, and the looming digital trust crisis.IN THIS EPISODE00:00 - Prompt Engineering, AI Agents & More03:27 - The Guests' Backgrounds in Cybersecurity and Trust & Safety09:31 - The Hidden Risks of Sharing Your Data with AI13:02 - Copyright vs. AI18:52 - The Digital Trust Crisis24:36 - Watermarking and Digital Verification30:25 - Using Proprietary Code with AI Assistants36:56 - AI Agents42:37 - Who This Book Is For (and Who It's Not For)

We celebrate Thanksgiving with some light hearted banter, and exploring the question of whether "High Watermarking" can lead to disappointment (even if it's hypothetical). We hope you're enjoying time with your family and friends this week, but if you want your weekly dose of personal finance content, we're here for you!Send us a textSend your questions for upcoming show to checkyourbalances@outlook.com @checkyourbalances on Instagram

Show NotesAs artificial intelligence begins generating music from vast datasets of human art, a fundamental question emerges: who truly owns the sound of AI? This episode of Music Evolves brings together a law student and former musician Chandler Lawn, music industry executive and professor Drew Thurlow, Michael Sheldrick, Co-Founder of Global Citizen, and intellectual property attorney Puya Partow-Navid, alongside hosts Sean Martin and Marco Ciappelli, to examine how AI is reshaping authorship, licensing, and the meaning of originality.The panel explores how AI democratizes creation while exposing deep ethical and economic gaps. Lawn raises the issue of whether artists whose works trained AI models deserve compensation, asking if innovation can be ethical when built on uncompensated labor. Thurlow highlights how, despite fears of automation, generative AI music accounts for less than 1% of streaming royalties—suggesting opportunity, not replacement.Sheldrick connects the conversation to a broader global context, describing how music's economic potential could drive sustainable development if nations modernize copyright frameworks. He views this shift as a rare chance to position creative industries as engines for jobs and growth.Partow-Navid grounds the discussion in legal precedent, pointing to landmark cases—from Two Live Crew to George R. R. Martin—as markers of how courts may interpret fair use, causality, and global jurisdiction in AI-driven creation.Together, the guests agree that the debate extends beyond legality. It's about the emotional authenticity that makes music human. As Chandler notes, “We connect through imperfection.” Marco adds that live performance may ultimately anchor value in a world saturated by digital replication.This conversation captures the tension—and promise—of a future where music, technology, and law must learn to play in harmony.GuestsChandler Lawn, AI Innovation and Law Fellow at The University of Texas School of Law | On LinkedIn: https://www.linkedin.com/in/chandlerlawn/Drew Thurlow, Adjunct Professor at Berklee College of Music | On LinkedIn: https://www.linkedin.com/in/drewthurlow/Michael Sheldrick, Co-Founder and Chief Policy, Impact and Government Affairs Officer at Global Citizen | On LinkedIn: https://www.linkedin.com/in/michael-sheldrick-30364051/Puya Partow-Navid, Partner at Seyfarth Shaw LLP | On LinkedIn: https://www.linkedin.com/in/puyapartow/Marco Ciappelli, Co-Founder, ITSPmagazine and Studio C60 | Website: https://www.marcociappelli.comHostSean Martin, Co-Founder at ITSPmagazine, Studio C60, and Host of Redefining CyberSecurity Podcast & Music Evolves Podcast | Website: https://www.seanmartin.com/ResourcesLegal Publication: You Can't Alway Get What You Want: A Survey of AI-related Copyright Considerations for the Music Industry published in Vol. 32, No. 3 of the Texas State Bar Entertainment and Sports Law Journal.BOOK: Machine Music: How AI Is Transforming Music's Next Act by Drew Thurlow: https://www.routledge.com/Machine-Music-How-AI-is-Transforming-Musics-Next-Act/Thurlow/p/book/9781032425242BOOK: From Ideas to Impact: A Playbook for Influencing and Implementing Change in a Divided World by Michael Sheldrick: https://www.fromideastoimpact.com/AI and Copyright Blogs:https://www.gadgetsgigabytesandgoodwill.com/category/ai/https://www.gadgetsgigabytesandgoodwill.com/2025/11/dr-thaler-is-right-in-part/https://www.gadgetsgigabytesandgoodwill.com/2025/07/californias-ai-law-has-set-rules-for-generative-ai-are-you-ready/https://www.gadgetsgigabytesandgoodwill.com/2025/06/copyright-office-firings-spark-constitutional-concerns-amid-ai-policy-tensions/Newsletter (Article, Video, Podcast): The Human Touch in a Synthetic Age: Why AI-Created Music Raises More Than Just Eyebrows: https://www.linkedin.com/pulse/human-touch-synthetic-age-why-ai-created-music-raises-martin-cissp-s9m7e/Article — Universal and Sony Music partner with new platform to detect AI music copyright theft using ‘groundbreaking neural fingerprinting' technology: https://www.musicbusinessworldwide.com/universal-and-sony-music-partner-with-new-platform-to-detect-ai-music-copyright-theft-using-groundbreaking-neural-fingerprinting-technology/Article: When Virtual Reality Is A Commodity, Will True Reality Come At A Premium: https://sean-martin.medium.com/when-virtual-reality-is-a-commodity-will-true-reality-come-at-a-premium-4a97bccb4d72Global Citizen: https://www.globalcitizen.org/Gallo Music (Gallo Records, South Africa): https://www.gallo.co.za/Global Citizen Festival: https://www.globalcitizen.org/en/festival/Andy Warhol Foundation v. Goldsmith (Shepard Fairey / “Hope” poster context): https://supreme.justia.com/cases/federal/us/598/21-869/case.pdfGeorge R. R. Martin / Authors Guild v. OpenAI (current AI training lawsuit): https://authorsguild.org/news/ag-and-authors-file-class-action-suit-against-openai/Campbell v. Acuff-Rose Music, Inc. (2 Live Crew “Pretty Woman”): https://supreme.justia.com/cases/federal/us/510/569/Vanilla Ice / “Under Pressure” Sampling Case: https://blogs.law.gwu.edu/mcir/case/queen-david-bowie-v-vanilla-ice/MIDiA Research — AI in Music Reports: https://www.midiaresearch.com/reports/ai-and-the-future-of-music-the-future-is-already-hereMerlin (Global Independent Rights Organization): https://www.merlinnetwork.org/Instagram Reel re: Spotify Terms: https://www.instagram.com/reel/DOrgbUNCYj_/ Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

Robert Bateman is a Senior Partner at Privacy Partnership, which provides consultancy and training on data protection and AI regulation, as well as legal advice via its associated law firm, Privacy Partnership Law. He also hosts The Privacy Partnership Podcast.This is Robert's third appearance on the show. We have covered three hot topics:* How far do we take watermarking of AI-generated content under article 50 of the AI Act?* How do pre-defined legitimate interest scenarios work under the UK Data (Use and Access) Act?* What is the tension between the Online Safety Act and the new data protection framework in the UK?References:SIGN UP NOW for the Masters of Privacy NYC LIVE recording and networking event on Nov 6 (if you happen to be in town)* Robert Bateman on LinkedIn* Robert Bateman on Bluesky* The Privacy Partnership Podcast* AI Act (EU Commission's resources)* Data (Use and Access) Act 2025: data protection and privacy changes* The EU approach to age verification (EU Commission)* EU follows UK with age verification in 2026 (PPC Land)* Wikipedia loses challenge against Online Safety Act verification rules (BBC)* Robert Bateman: the EDPB's Opinion on auditing subprocessors and the future of Meta's unskippable ads (Masters of Privacy, Nov 2024)* Robert Bateman: Consent or Pay (Masters of Privacy, Oct 2023) This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.mastersofprivacy.com/subscribe

In this episode, Anna Rose and Tarun Chitra chat with Miranda Christ, a computer science PhD student at Columbia University, about the intersection of cryptography and AI through watermarking techniques. Miranda shares her research on developing imperceptible ways to prove that content was created by AI models, covering everything from simple red-green word lists to sophisticated pseudorandom error-correcting codes. The discussion explores the cryptographic properties of watermarks - including completeness, soundness, and undetectability - and how these parallel the properties we see in zero-knowledge proof systems. Miranda explains how watermarking differs from other cryptographic approaches like ZKML by only modifying the sampling process rather than the underlying model weights, making it computationally lightweight and practical for deployment. Related links: Episode 206: Distilling DeFi Primitives with Guillermo, Alex and Tarun My AI Safety Lecture for UT Effective Altruism Google SynthID Amazon Public Watermark Detector How ChatGPT could embed a ‘watermark' in the text it generates - New York Times Wall Street Journal on OpenAI not Deploying Watermarks A Watermark for Large Language Models Undetectable Watermarks for Language Models Watermarks in the Sand: Impossibility of Strong Watermarking for Generative Models Pseudorandom Error-Correcting Codes Ideal Pseudorandom Codes Check out the latest jobs in ZK at the ZK Podcast Jobs Board

In this AI research paper reading, we dive into "A Watermark for Large Language Models" with the paper's author John Kirchenbauer. This paper is a timely exploration of techniques for embedding invisible but detectable signals in AI-generated text. These watermarking strategies aim to help mitigate misuse of large language models by making machine-generated content distinguishable from human writing, without sacrificing text quality or requiring access to the model's internals.Learn more about the A Watermark for Large Language Models paper. Learn more about agent observability and LLM observability, join the Arize AI Slack community or get the latest on LinkedIn and X.Learn more about AI observability and evaluation, join the Arize AI Slack community or get the latest on LinkedIn and X.

Sometimes it's easy to tell whether a video is fake, other times, it's not. Watermarking is used to digitally stamp fake videos, whether that stamp is visible to the human eye or is embedded in the video's data. But with new technology that allows for the stamp to be removed without anyone noticing, how is regulation enforced? Host Mike Eppel speaks to Andre Kassis, University of Waterloo PhD candidate in computer science, and Angus Lockhart, senior policy analyst at 'The Dais' with Toronto Metropolitan University to discuss the safeguards in place to ensure AI-produced content is labelled accordingly and who can be held accountable if the rules start to bend. We love feedback at The Big Story, as well as suggestions for future episodes. You can find us:Through email at hello@thebigstorypodcast.ca Or @thebigstoryfpn on Twitter

When I reflect on all the exciting things happening with the podcast, not only am I incredibly grateful, but it's also really fun to look back at how it all started. So in this episode, I'm re-listening to my first handful of episodes and asking myself, “Do they still hold up?!” I thought before listening that I had a pretty good idea of what I said in these early episodes, but I actually kind of surprised myself sometimes! Listen in as I share the details on what has stood the test of time… and what I have a new way of thinking about. Here's a preview: Stood the test of time: 3 important things for your Instagram Email signature Directing people to a contact form on your website Getting a Google Business Profile Instagram caption ideas & social media schedulers Automation ideas & endorsement for 17hats CRM Profit First cash flow system I have a new way of thinking about: Some of the resources I created and mentioned in the early episodes Business email accounts Importance of styled shoots Watermarking photos   I also share my episode plan for the year, in case you're curious! And in the UGlu Hotline, hear a tip for transporting framework in a personal vehicle.   RESOURCES MENTIONED: Presenting sponsor: 17hats (get 50% off your 1st year)  Other sponsors & resources: Havin' A Party Wholesale (save 5% with code BRIGHT) Courtney Lynette Creative Co 2025 Bright Balloon Business Planner  UGlu by Pro Tapes (save 5% at Havin' A Party with code BRIGHT)  Call into the UGlu Hotline to ask a question or leave advice! (262) 221-8514 Balloon Boss Mastermind & Summit  - - - - Get bonus episodes 50 Ideas for Email Marketing | Join the Bright Balloon email list  Freeform Freedom | More courses  @thebrightballoon  The Bright Balloon on YouTube 

How fast is the AI race really going? What is the current state of Quantum Computing? What actually *is* the P vs NP problem? - former OpenAI researcher and theoretical computer scientist Scott Aaronson joins Liv and Igor to discuss everything quantum, AI and consciousness. We hear about his experience working on OpenAI's "superalignment team", whether quantum computers might break Bitcoin, the state of University Admissions, and even a proposal for a new religion! Strap in for a fascinating conversation that bridges deep theory with pressing real-world concerns about our technological future. Chapters: 1:30 - Working at OpenAI 4:23 - His Approaches to AI Alignment 6:23 - Watermarking & Detection of AI content 19:15 - P vs. NP 27:11 - The Current State of AI Safety 37:38 - Bad "Just-a-ism" Arguments around LLMs 48:25 - What Sets Human Creativity Apart from AI 55:30 - A Religion for AGI? 1:00:49 - More Moral Philosophy 1:05:24 - The AI Arms Race 1:11:08 - The Government Intervention Dilemma 1:23:28 - The Current State of Quantum Computing 1:36:25 - Will QC destroy Cryptography? 1:48:55 - Politics on College Campuses 2:03:11 - Scott's Childhood & Relationship with Competition 2:23:25 - Rapid-fire Predictions Links: ♾️ Scott's Blog: ⁠https://scottaaronson.blog/⁠ ♾️ Scott's Book: ⁠https://www.amazon.com/Quantum-Computing-since-Democritus-Aaronson/dp/0521199565⁠ ♾️ QIC at UTA: https://www.cs.utexas.edu/~qic/ Credits Credits: ♾️  Hosted by Liv Boeree and Igor Kurganov ♾️  Produced by Liv Boeree ♾️  Post-Production by Ryan Kessler The Win-Win Podcast: Poker champion Liv Boeree takes to the interview chair to tease apart the complexities of one of the most fundamental parts of human nature: competition. Liv is joined by top philosophers, gamers, artists, technologists, CEOs, scientists, athletes and more to understand how competition manifests in their world, and how to change seemingly win-lose games into Win-Wins. #WinWinPodcast #QuantumComputing #AISafety #LLM

Agents, agents, and more agents! In Episode 27 of Mixture of Experts, host Tim Hwang is joined by Volkmar Uhlig and Vyoma Gajjar. First, the experts chat about Mark Benioff's spicy tweet, and what this means for the future of AI agents. Next, how much energy is needed to power AI models, and should we be concerned? Then, the experts debrief Anthropic's release of computer use. Finally, Google is integrating SynthID-Text into Gemini to help watermark AI-generated text, do we need this feature? Learn more on today's Mixture of Experts.The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity.

NTD Good Morning—8/26/20241. US Works to Avert Escalation After Israel-Hezbollah Exchange2. Reuters Safety Adviser Killed in Missile Strike in Ukraine3. US Targets Russian, Chinese Firms for Aiding Russia4. Harris, Walz Head to Georgia5. Watermarking, Validation Needed to Combat AI Risk: Advocate6. Flash Flooding in Grand Canyon Claims Life of Hiker7. French Media: Telegram Co-Founder Detained8. Microsoft Plans Summit After Crowdstrike Outage9. Italy Begins Manslaughter Probe Into Yacht Sinking10. SpaceX Will Bring 2 NASA Astronauts Home11. How Will Kennedy Endorsement Affect Trump Campaign?12. Dr. Fauci Recovering From West Nile Virus13. Why Teens are so Drawn to Social Media14. How Travel Spending is Shifting15. NJ Transit Offering 'Fare Holiday' Next Week16. French Mark 80 Years of Paris Liberation From Nazis17. Paralympic Torch Crosses Channel and Begins French Journey18. Titanic Article Discovered in Wardrobe Auctioned Off19. Redheads Gather to Light up Dutch Festival20. Chef Defies Doctors With Device That Keeps Him Cooking21. The Resilient Academic: Barbara Gitenstein's Journey22. What's Next for the Democratic and Republican Parties?23. Hurricane Sales Tax Holiday Underway in Florida24. Mountains See Early Dusting of Snow in August25. 102-Year-Old Woman Becomes UK's Oldest Skydiver26. Why Inflation is Low But Prices Remain High27. Realtors Adapt to New Buyer Agent Rules

In episode 578, Kathy Berget teaches us what to do when our blog images have been used without our permission and how to get compensated for them. Kathy is the author, photographer, recipe developer, and writer at Beyond the Chicken Coop where she creates delicious home-cooked recipes utilizing what they grow and raise. Kathy is a former elementary school principal and has three grown kids; twin boys and one girl. Kathy and her husband live in the country on their own little farm. In this episode, you'll learn about copyright for images, ways in which you can respond and how to use image theft protection services like Pixsy to get compensated. Key points discussed: - Image theft is a common issue for food bloggers: Food blog images are often stolen or used without permission by various businesses, including restaurants, markets, and online retailers. - Copyright protection is automatic for blog images: Photographers automatically have copyright over their images, even without formal registration. - Services like Pixsy can help track and fight image theft: Pixsy is an online service that helps photographers find and fight unauthorized use of their images. - Responding to image theft requires a balanced approach: Do not obsess over every instance of image theft -  selectively pursue cases that are worth the time and effort (especially if images are used for commercial purposes). - Watermarking images can increase their value if stolen: Removing or altering a watermark on an image can actually increase its value if used without permission. - Educating the public about image rights is an ongoing challenge: Many people may be unaware that using images found online without permission is considered theft. - Persistence and documentation are key when pursuing image theft cases: It's important to thoroughly document evidence and follow through with service providers like Pixsy. - Maintaining a positive attitude is important when dealing with image theft: Do no let image theft issues negatively impact your overall mindset and productivity. If You Loved This Episode… You'll love Episode 390 with Rob Finkelstein - Legal Issues Every Food Photographer Should Consider Connect with Kathy Berget Website | Instagram

CrowdStrike said Delta's woes aren't its fault after the massive IT outage, OpenAI confirms it's looking into text watermarking for ChatGPT that could expose cheating students, and Nat Geo's first Vision Pro immersive environment takes you to Iceland. It's Tuesday, August 6th and this is Engadget News. Learn more about your ad choices. Visit podcastchoices.com/adchoices

Plus, Apple has finally started sending out payments from its butterfly keyboard settlement. Learn more about your ad choices. Visit podcastchoices.com/adchoices

#SecurityConfidential #DarkRhiinoSecurity Aaron is a Security Confidential Alumni, Entrepreneur, Author, former VP of Microsoft in China, and the CEO of Nametag Inc, the company that invented “Sign in with ID” as a more secure alternative to passwords. 00:00 Intro 00:57 Our Guest 01:46 Social Engineering trends 04:03 Deep fakes: how does it work? 09:18 Watermarking content 11:30 Deepfake Prevention: Injection attack 13:11: Deepfake prevention: Presentation attack 15:00 How do you verify behind a screen? 27:16 Hidden security in your phones 32:08 Social Engineering and MFA in Healthcare 41:18 How to maintain LOYAL Employees 46:15 China: Friend or Foe? 50:13 Connecting with Aaron ------------------------------------------------------------------ Watch our other episode with Aaron: https://youtu.be/m2PLow9cWSE ------------------------------------------------------------------ To learn more about Nametag visit https://getnametag.com/ To learn more about Dark Rhiino Security visit https://www.darkrhiinosecurity.com ----------------------------------------------------------------- SOCIAL MEDIA: Stay connected with us on our social media pages where we'll give you snippets, alerts for new podcasts, and even behind the scenes of our studio! Instagram: @securityconfidential and @Darkrhiinosecurity Facebook: @Dark-Rhiino-Security-Inc Twitter: @darkrhiinosec LinkedIn: @dark-rhiino-security Youtube: @DarkRhiinoSecurity ​ ------------------------------------------------------------------ #darkrhiinosecurity #securityconfidential #cybersecurity #cyberpodcast #ai #artificialintelligence #securitypodcast #cybernews #technews #techsoftware #informationtechnology #infosec #cybersecurityforbeginners #technewstoday

In this episode, we discuss Microsoft's investment in G42 and questions surrounding G42's ties to China (1:12), the latest reporting about the Israeli military's use of AI and policy implications advanced technologies in warfare (9:23), and Meta's new watermarking policy (23:01). aipolicypodcast@csis.org Wadhwani Center for AI and Advanced Technologies | CSIS The DARPA Perspective on AI and Autonomy at the DOD | CSIS Events Scaling AI-enabled Capabilities at the DOD: Government and Industry Perspectives: The State of DOD AI and Autonomy Policy:

Dive into the realm of innovation with Microsoft's revolutionary AI features and their strategy for implementing image watermarking. Explore how these advancements are poised to reshape the AI landscape and enhance content protection in the digital era. Get on the AI Box Waitlist: AIBox.aiJoin our ChatGPT Community: Facebook GroupFollow me on Twitter: Jaeden's Twitter

In this episode, we analyze Steg.AI's recent $5 million seed round, exploring the significance of their invisible watermarking technology and its potential applications in enhancing the security of images and documents. Invest in AI Box: ⁠https://Republic.com/ai-box⁠ Get on the AI Box Waitlist: ⁠https://AIBox.ai/⁠ ⁠AI Facebook Community Learn About ChatGPT Learn About AI at Tesla

Embark on a journey into the realm of invisible safeguards as DeepMind and Google Cloud collaborate on AI image watermarking innovation. Explore the protective measures that redefine the security of visual content. Get on the AI Box Waitlist: AIBox.ai Join our ChatGPT Community: Facebook Group Follow me on Twitter: Jaeden's Twitter

In this episode, we explore the world of invisible signatures as I delve into the confluence between DeepMind and Google Cloud in the arena of AI image watermarking. Join me for a solo discussion, where we uncover the techniques, applications, and the transformative potential of this collaboration in safeguarding digital imagery. Invest in AI Box: ⁠https://Republic.com/ai-box⁠ Get on the AI Box Waitlist: ⁠https://AIBox.ai/⁠ ⁠AI Facebook Community Learn About ChatGPT Learn About AI at Tesla

On this episode of The AI Moment, we discuss an emerging Gen AI trend: Watermarking & other strategies for licensing AI training data & combating malicious AI generated content. As we move into year two of Generative AI, some themes have emerged in terms of the downsides to the technology.  Two of the biggest downsides have been: Combating malicious or misleading AI-generated content, and  Copyright/IP rights for both non AI generated and AI generated content  Initiatives by Google, Fox-Polygon, the Content Authenticity Initiative and academic researchers focused primarily on digital watermarking are the latest and most prominent attempts to address these issues. What will the impact of this trend be to enabling or stifling gen AI be?

2024 will be the biggest election year in world history. Forty countries will hold national elections, with over two billion voters heading to the polls. In this episode of Your Undivided Attention, two experts give us a situation report on how AI will increase the risks to our elections and our democracies. Correction: Tristan says two billion people from 70 countries will be undergoing democratic elections in 2024. The number expands to 70 when non-national elections are factored in.RECOMMENDED MEDIA White House AI Executive Order Takes On Complexity of Content Integrity IssuesRenee DiResta's piece in Tech Policy Press about content integrity within President Biden's AI executive orderThe Stanford Internet ObservatoryA cross-disciplinary program of research, teaching and policy engagement for the study of abuse in current information technologies, with a focus on social mediaDemosBritain's leading cross-party think tankInvisible Rulers: The People Who Turn Lies into Reality by Renee DiRestaPre-order Renee's upcoming book that's landing on shelves June 11, 2024RECOMMENDED YUA EPISODESThe Spin Doctors Are In with Renee DiRestaFrom Russia with Likes Part 1 with Renee DiRestaFrom Russia with Likes Part 2 with Renee DiRestaEsther Perel on Artificial IntimacyThe AI DilemmaA Conversation with Facebook Whistleblower Frances HaugenYour Undivided Attention is produced by the Center for Humane Technology. Follow us on Twitter: @HumaneTech_ 

Content authenticity and enforcing copyright in the age of AI are proving difficult problems to solve.

In this episode, we discuss the AI gold rush and its impact on businesses. According to a study by IDC, companies are reaping 3.5 times returns on their AI investments, with a return on investment within 14 months on average. The report also highlights how generative AI is driving increased interest and investment in the technology. Additionally, the Oxford Internet Institute conducted a study that found AI skills and knowledge can increase a worker's salary by up to 40%. The study examined over 1,000 skills in 25,000 workers, showing the positive impact of AI-related knowledge on potential salaries.Three things to know today00:00 The AI Gold Rush Pays: Companies Reap 3.5x Returns & Salaries Surge by 40%04:50 Federal AI Blueprint Draws Industry Eyeballs as OMB Solicits Public Wisdom07:58 Watermarking, the AI Apocalypse, & Adult Content Leveraging AIAdvertiser: https://movebot.io/Looking for a link from the stories? The entire script of the show, with links to articles, are posted in each story on https://www.businessof.tech/Do you want the show on your podcast app or the written versions of the stories? Subscribe to the Business of Tech: https://www.businessof.tech/subscribe/Support the show on Patreon: https://patreon.com/mspradio/Want our stuff? Cool Merch? Wear “Why Do We Care?” - Visit https://mspradio.myspreadshop.comFollow us on:LinkedIn: https://www.linkedin.com/company/28908079/YouTube: https://youtube.com/mspradio/Facebook: https://www.facebook.com/mspradionews/Instagram: https://www.instagram.com/mspradio/TikTok: https://www.tiktok.com/@businessoftech

In this episode, we delve into the groundbreaking $5 million seed funding round acquired by Steg.AI, a trailblazing company specializing in invisible watermarking for images and documents. Join us as we explore the implications of this significant investment and how Steg.AI is poised to transform the way we secure and protect digital content. Discover the innovative technology behind invisible watermarking and its potential applications in various industries. Get on the AI Box Waitlist: https://AIBox.ai/Join our ChatGPT Community: ⁠https://www.facebook.com/groups/739308654562189/⁠Follow me on Twitter: ⁠https://twitter.com/jaeden_ai⁠

This episode is sponsored by Shopify. Shopify is a commerce platform that allows anyone to set up an online store and sell their products. Whether you're selling online, on social media, or in person, Shopify has you covered on every base. With Shopify you can sell physical and digital products. You can sell services, memberships, ticketed events, rentals and even classes and lessons. Sign up for a $1 per month trial period at http://shopify.com/eyeonai On episode #145 of Eye on AI, Craig Smith sits down with Riley McCormack, President and CEO of Digimarc, pioneers in digital watermarking and cloud-based product data. In this episode, we explore the critical role of digital watermarking in securing our digital assets, especially amidst the surging influence of AI. Riley guides us through the potential risks and benefits this technology brings to the forefront as AI continues to transform our digital world. We then navigate the intricate territories of NFTs and distributed ledger technology, understanding how digital watermarking is reshaping these fields by ensuring trust and authenticity. Our discussion also delves into the Digital Millennium Copyright Act of 1997, highlighting its relevance in upholding copyrights and fostering trust within the digital stratosphere. We conclude with a look at how digital watermarking impacts content creation and its role in shaping a secure and sustainable digital future, from collaborations with central banks to innovative products like Digimarc Recycle.   Craig Smith Twitter: https://twitter.com/craigss Eye on A.I. Twitter: https://twitter.com/EyeOn_AI   00:00 Preview, Introduction and Shopify 03:05 Introduction to Digital Watermarking  07:01 Evolution of Digital Watermarking  14:21 Digimarc's Role in Digital Watermarking 21:13 Exploring the Protection Of Digital Content  28:56 Key Characteristics of Digital Watermarking  35:12 Application and Implementation of Digital Watermarking  42:46 Watermarking in Blockchain  49:11 What is Digimarc Validate? 01:03:01 Outro and Shopify

In this episode, we explore DeepMind's recent partnership with Google Cloud to implement watermarking on AI-generated images, aiming to increase transparency and address ethical concerns. We'll dive into the technology behind watermarking, how this move could impact the AI and digital art communities, and why this is a significant step toward responsible AI use. Get on the AI Box Waitlist: ⁠https://AIBox.ai/⁠ Facebook Community: ⁠⁠https://www.facebook.com/groups/739308654562189/⁠⁠ Discord Community: ⁠https://aibox.ai/discord⁠ Follow me on X: ⁠https://twitter.com/jaeden_ai⁠

In this episode, we delve into Steg.AI's recently secured $5M seed funding, focusing on their innovative technology for invisible watermarking on images and documents. We discuss the company's mission, the problem they're solving in digital security, and the potential impact of their unique solution in the era of rampant data breaches and intellectual property theft. Get on the AI Box Waitlist: ⁠https://AIBox.ai/⁠ Investor Contact Email: jaeden@aibox.ai Facebook Community: ⁠⁠https://www.facebook.com/groups/739308654562189/⁠⁠ Discord Community: ⁠https://aibox.ai/discord⁠ Download Selfpause: ⁠https://selfpause.com/Podcast⁠ Follow me on Twitter... er... X.com: ⁠https://twitter.com/jaeden_ai⁠

Today we're joined by Tom Goldstein, an associate professor at the University of Maryland. Tom's research sits at the intersection of ML and optimization and has previously been featured in the New Yorker for his work on invisibility cloaks, clothing that can evade object detection. In our conversation, we focus on his more recent research on watermarking LLM output. We explore the motivations behind adding these watermarks, how they work, and different ways a watermark could be deployed, as well as political and economic incentive structures around the adoption of watermarking and future directions for that line of work. We also discuss Tom's research into data leakage, particularly in stable diffusion models, work that is analogous to recent guest Nicholas Carlini's research into LLM data extraction.