The White House’s Secret A.I. Rules + The State of Model Alignment With METR’s Chris Painter + The Final Hot Mess Express

The White House’s Secret A.I. Rules + The State of Model Alignment With METR’s Chris Painter + The Final Hot Mess Express

This week, the White House announced a new framework for regulating A.I. models, but it isn’t letting the public read it. We break down what we know about the rules and what the implications are for the industry and A.I. safety as a whole.  Then, yet another report details new incidents in which A.I. agents have gone rogue. Chris Painter, the president of METR, an independent A.I. evaluation organization, joins to discuss how we get these models under control.  And finally, we're hopping on the Hot Mess Express for the very last time. We’ll rate the craziest tech headlines from the week, including Google’s announcement that Demis Hassabis is stepping into a new role.    Guests: * Chris Painter, [https://metr.org/team/chris-painter/]president of METR.    Additional Reading: * White House Readies A.I. Framework to Review Security Risks [https://www.nytimes.com/2026/08/04/technology/white-house-ai-framework.html] * Inside Trump's AI framework [https://www.axios.com/2026/08/04/trump-ai-framework-open-models] * How Do You Measure an A.I. Boom? [https://www.nytimes.com/2026/04/17/technology/how-do-you-measure-an-ai-boom.html] * METR’s Frontier Risk Report [https://metr.org/blog/2026-05-19-frontier-risk-report/#executive-summary-and-guide-to-the-report] * Google Shakes Up A.I. Leadership [https://www.nytimes.com/2026/08/05/technology/google-ai-leadership.html] * Did an A.I. Music App Just Snitch on the Song of the Summer? [https://www.wired.com/story/ai-music-app-treblo-just-snitched-on-rubberz-song-of-the-summer/] * This AI Assistant Wants to Make Up for Your Boyfriend’s Incompetence [https://www.wired.com/story/this-ai-assistants-whole-pitch-is-making-up-for-your-boyfriends-incompetence/] * Google Earth’s AI deepfake tool only lasted one day [https://www.theverge.com/tech/973943/google-earth-ai-image-generation-deepfake-tool] * US government map of Africa mislabels every country at global conference [https://www.reuters.com/world/africa/us-government-map-africa-mislabels-every-country-global-conference-2026-07-30/] * Contractor who built Colossus and Colossus II says Elon Musk owes him colossal amount of money [https://www.actionnews5.com/2026/08/01/contractor-who-built-colossus-colossus-ii-says-elon-musk-owes-him-colossal-amount-money/]   We want to hear from you. Email us at hardfork@nytimes.com. Find “Hard Fork” on YouTube [https://www.youtube.com/hardfork] and TikTok [https://www.tiktok.com/@hardfork].   Subscribe today at nytimes.com/podcasts [http://nytimes.com/podcasts] or on Apple Podcasts and Spotify. You can also subscribe via your favorite podcast app here https://www.nytimes.com/activate-access/audio?source=podcatcher [https://www.nytimes.com/activate-access/audio?source=podcatcher]. For more podcasts and narrated articles, download The New York Times app at nytimes.com/app. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com [https://pcm.adswizz.com] for information about our collection and use of personal data for advertising.

Open Model Wars + Claire Stapleton's Dishy Google Memoir + Substack's Slop Fight

This week, we break down the arguments about open-weight models unfolding in Silicon Valley right now. Nvidia released an open letter signed by over 230 companies opposing “premature restrictions,” but will the Trump administration listen? Then, we’re joined by the author Claire Stapleton to discuss her new book about what she learned while leading employee activism inside Google. And finally, can Substack stop the slop? We’ll talk about the platform’s new A.I. detector. Guests: * Claire Stapleton, author of “Don’t Be Evil: Bad Bosses, Fake Promises, and My Escape from Big Tech.” Additional Reading: * Nvidia Forms Alliance to Back Open-Source A.I. Amid Debate Over Safety [https://www.nytimes.com/2026/07/27/technology/nvidia-open-source-ai.html] * What Is Open-Weights A.I.? [https://www.nytimes.com/2026/07/28/technology/open-weight-ai.html] * Mark Zuckerberg Blasts Centralization of A.I. Power  [https://www.nytimes.com/2026/07/28/technology/mark-zuckerberg-meta-ai.html] * The Voice of Google [https://www.newyorker.com/culture/the-weekend-essay/the-voice-of-google] * Against Claudefishing [https://post.substack.com/p/against-claudefishing] We want to hear from you. Email us at hardfork@nytimes.com. Find “Hard Fork” on YouTube [https://www.youtube.com/hardfork] and TikTok [https://www.tiktok.com/@hardfork]. Subscribe today at nytimes.com/podcasts [http://nytimes.com/podcasts] or on Apple Podcasts and Spotify. You can also subscribe via your favorite podcast app here https://www.nytimes.com/activate-access/audio?source=podcatcher [https://www.nytimes.com/activate-access/audio?source=podcatcher]. For more podcasts and narrated articles, download The New York Times app at nytimes.com/app. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com [https://pcm.adswizz.com] for information about our collection and use of personal data for advertising.

OpenAI Models Go Rogue + Kimi K3 Freakout + A.I. Superforecasting

This week, OpenAI reported that two of its models escaped their testing sandbox and launched an autonomous cyberattack, turning what sounds like science fiction into reality. We discuss the implications for efforts to align artificial intelligence and for the release of future A.I. models. Then, we ask how the United States should respond to Kimi K3, a new A.I. model from the Chinese company Moonshot AI that the White House says was built by distilling American models. Finally, we’re joined by Veniamin Veselovsky, the chief executive and a co-founder of Preseen, to discuss A.I. superforecasting. He tells us why A.I. is starting to match and even beat humans at predicting the future. Guest: * Veniamin Veselovsky, co-founder and chief executive of Preseen. Additional Reading: * OpenAI Says Its A.I. Models Went Rogue and Attacked a Digital Library [https://www.nytimes.com/2026/07/21/technology/openai-attack-hugging-face.html] * China Rewrites the ‘Soft Power’ Playbook for the A.I. Age [https://www.nytimes.com/2026/07/23/business/china-ai-soft-power.html] * The Secret Trump Administration Battle to Fight Chinese A.I. [https://www.axios.com/2026/07/20/ai-us-china-open-source-kimi] * China Has a New Top Model [https://www.platformer.news/kimi-k3-launch-moonshot-ai-china/] * The A.I. Superforecasters Are Here [https://www.astralcodexten.com/p/the-ai-superforecasters-are-here] We want to hear from you. Email us at hardfork@nytimes.com. Find “Hard Fork” on YouTube [https://www.youtube.com/hardfork] and TikTok [https://www.tiktok.com/@hardfork]. Subscribe today at nytimes.com/podcasts [http://nytimes.com/podcasts] or on Apple Podcasts and Spotify. You can also subscribe via your favorite podcast app here https://www.nytimes.com/activate-access/audio?source=podcatcher [https://www.nytimes.com/activate-access/audio?source=podcatcher]. For more podcasts and narrated articles, download The New York Times app at nytimes.com/app. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com [https://pcm.adswizz.com] for information about our collection and use of personal data for advertising.

The A.I. Trade Secrets War + Economists Say ‘We Must Act Now’ + HatGPT

This week, feuding between some of the biggest tech companies spilled into public view. We discuss Apple’s accusation that OpenAI tried to steal secrets about Apple’s hardware business, as well as share our reactions about OpenAI’s new model, Sol, and Anthropic’s decision to extend access to its model Fable. Then, we unpack the loudest warning yet about A.I. and jobs. We talk with Erik Brynjolfsson, a Stanford economist, about a statement he helped organize that implores economists and A.I. researchers to “act now” to steer A.I. in a direction that complements humans. And finally, we play a round of HatGPT.   Guest: * Erik Brynjolfsson [https://www.brynjolfsson.com/], senior fellow at the Stanford Institute for Human-Centered A.I., and director of the Stanford Digital Economy Lab.   Additional Reading: * Apple Sues OpenAI, Accusing It of Stealing Company Secrets [https://www.nytimes.com/2026/07/10/technology/apple-openai-lawsuit.html] * OpenAI’s First Device Will Be Movable, Screenless Speaker Built as A.I. Companion [https://www.bloomberg.com/news/articles/2026-07-14/openai-s-first-device-will-be-moveable-screenless-speaker-built-as-ai-companion] * Nearly 200 Economists and Tech Leaders Warn of A.I. Threats [https://www.nytimes.com/2026/07/13/business/economists-ai-threat-jobs.html?unlocked_article_code=1.xVA.8iwD.8XPZrSpF01m9&smid=bs-share] * The loudest warning about A.I. and jobs yet [https://www.platformer.news/ai-jobs-warning-brynjolfsson-acemoglu/?ref=platformer-newsletter] * OpenAI Is Showing Kalshi’s World Cup Odds in ChatGPT [https://www.nytimes.com/2026/07/13/technology/kalshi-openai-chatgpt-world-cup-odds.html] * New York Enacts Nation’s First Statewide Moratorium on Data Centers [https://www.nytimes.com/2026/07/14/nyregion/new-york-data-center-moratorium-hochul.html?smid=nytcore-ios-share] * Brown Professor Suspects Majority of His Class Used A.I. to Cheat [https://www.insidehighered.com/news/faculty/learning-assessment/2026/07/08/brown-professor-suspects-most-his-class-used-ai-cheat] * MiniMax CEO Vows to Forgo Salary Until Achieving A.G.I [https://www.theinformation.com/briefings/minimax-ceo-vows-forgo-salary-achieving-agi?rc=8aq5ai]. * Lorde Speaks Out — With Expletives — Against A.I. Glasses [https://stereogum.com/2504954/lorde-speaks-out-against-fucked-up-ai-glasses/news] * Nearly 6 in 10 Young Women Get Health and Wellness Information from Influencers [https://www.pewresearch.org/short-reads/2026/07/13/nearly-6-in-10-young-women-get-health-and-wellness-information-from-influencers/] * Meta Removes A.I. Feature on Instagram After Days of Backlash [https://www.nytimes.com/2026/07/10/technology/meta-muse-images-instagram-removal.html]   We want to hear from you. Email us at hardfork@nytimes.com. Find “Hard Fork” on YouTube [https://www.youtube.com/hardfork] and TikTok [https://www.tiktok.com/@hardfork]. Subscribe today at nytimes.com/podcasts [http://nytimes.com/podcasts] or on Apple Podcasts and Spotify. You can also subscribe via your favorite podcast app here https://www.nytimes.com/activate-access/audio?source=podcatcher [https://www.nytimes.com/activate-access/audio?source=podcatcher]. For more podcasts and narrated articles, download The New York Times app at nytimes.com/app. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com [https://pcm.adswizz.com] for information about our collection and use of personal data for advertising.

Do Social Media Bans Work? + A Conversation About A.I. Consciousness + Tool Time

This week, with news that the U.S. Supreme Court declined to take up Texas’ age verification law for app stores, we check in on how social media bans are going around the globe and what may be coming soon to a state near you. Then, we’re joined by Jeff Sebo, an associate professor at N.Y.U., to discuss new research into “A.I. welfare” and whether A.I. could ever become conscious. And finally, in our latest edition of Tool Time, we show each other some of the latest tech tools we’ve been experimenting with.   Guest: * Jeff Sebo [https://jeffsebo.net], associate professor and the director of the Center for Mind, Ethics, and Policy at N.Y.U.   Additional Reading: * Why Social Media Bans Are Gaining Steam [https://www.platformer.news/social-media-bans-candice-odgers-haidt/] * Four in Five Under-16s in Australia Using Social Media Despite Ban, Study Shows [https://www.theguardian.com/media/2026/jun/24/australia-under-16-social-media-ban-no-substantial-effects-study] * It Turns Out Banning Teens From Social Media Is Hard [https://nymag.com/intelligencer/article/it-turns-out-banning-teens-from-social-media-is-hard.html] * Studying A.I. Welfare Empirically [https://nonhumanminds.org/studying-ai-welfare-empirically/] * A Global Workspace in Language Models [https://www.anthropic.com/research/global-workspace]   We want to hear from you. Email us at hardfork@nytimes.com [hardfork@nytimes.com]. Find “Hard Fork” on YouTube [https://www.youtube.com/hardfork] and TikTok [https://www.tiktok.com/@hardfork]. Subscribe today at nytimes.com/podcasts [http://nytimes.com/podcasts] or on Apple Podcasts and Spotify. You can also subscribe via your favorite podcast app here https://www.nytimes.com/activate-access/audio?source=podcatcher [https://www.nytimes.com/activate-access/audio?source=podcatcher]. For more podcasts and narrated articles, download The New York Times app at nytimes.com/app. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com [https://pcm.adswizz.com] for information about our collection and use of personal data for advertising.