Doom Debates!

Brendan and Randan Steinhauser are a husband and wife team who know how to make things happen in conservative politics.

Brendan helped build the Tea Party movement and ran John Cornyn’s Senate campaign. Randan has been involved in advocacy and political campaigns in more than 30 states, including working under former Secretary of Education Betsy DeVos.

Now they’re bringing Republicans, parents, and faith leaders into the fight against AI catastrophe.

They know how to build political momentum, but my question is whether that will lead to action quickly enough. I'm skeptical of incremental legislation efforts and data center protests, but they argue it's necessary coalition-building that leads to superintelligence regulation.

This isn't your typical state-of-the-political-map podcast. We also get into:

Their personal AGI awakenings

How people of faith make sense of claims of Godlike AI and AI doom

My own transition to joining the Republican Party

Transhumanism, souls, and mind uploading

Parenting in the age of AI

It’s a candid conversation with a couple who are making the fight for our future feel a little more winnable. Enjoy!

Watch on YouTube: https://www.youtube.com/watch?v=XTm9tqS1Ibw

Timestamps

00:00:00 — Cold Open

00:01:07 — Introducing Brendan & Randan Steinhauser

00:02:22 — Why I Came Out as Republican

00:04:28 — The Alliance for Secure AI

00:06:30 — Losing Control of Superintelligence

00:07:52 — Brendan, What’s Your P(Doom)?™

00:08:26 — Randan, What’s Your P(Doom)?™

00:10:47 — Finding the AI Safety Movement

00:15:04 — Trump, Republicans, and Big Tech Money

00:19:22 — Bringing Faith Leaders Into AI Safety

00:23:23 — God’s Plan and Human Extinction

00:29:43 — Priorities for Congress

00:31:42 — The Morality of AI Companies

00:34:03 — PauseAI and Employee Whistleblowers

00:43:34 — Effective Altruism and a Broader Coalition

00:48:12 — Transhumanism, Souls, and Mind Uploading

00:57:40 — Building a Machine God

01:03:20 — Raising Children Under AI Risk

01:07:35 — Data Center Protests and Movement Building

01:13:09 — Talking to People About AI Risk

01:17:32 — State Laws and the 2028 Election

01:20:38 — Getting Involved

Links

REFERENCED IN THE EPISODE

AI Risk Evaluation Act — Hawley–Blumenthal bill text

The Singularity Is Near — Ray Kurzweil

Can We Stop the AI Apocalypse? — Dan Crenshaw interviews Eliezer Yudkowsky

Magnifica Humanitas — Pope Leo XIV’s encyclical on humanity in the age of AI

Pacing the Frontier — open letter from frontier AI employees

PauseAI — the global movement to pause unsafe AI development

PauseAI US — organizing and advocacy in the United States

Jacob Coxon’s resignation statement — the insider warning discussed

Americans for Responsible Innovation — Brad Carson’s bipartisan advocacy organization

PAST DOOM DEBATES EPISODES MENTIONED

Let’s CALL OUT the AI Doom “Enablers” — Holly Elmore on working at AI labs

God vs. AI Doom: Debate with Bentham’s Bulldog

DOOM DEBATES

Join the Doom Debates Discord

Support Doom Debates — become a mission partner

Doom Debates’ Mission is to raise mainstream awareness of imminent extinction from AGI and build the social infrastructure for high-quality debate.

Support the mission by subscribing to my Substack at DoomDebates.com and to youtube.com/@DoomDebates, or to really take things to the next level: Donate 🙏

Bestselling author Robert Wright and I discuss this week’s stock market jitters, Google’s new interactive world model, and why I sometimes spend $1,000 in a day on AI agents. Plus:

The Binance hack;

An AI audit of two star economists;

Anthropic’s financial prospects;

And whether AI could crack the math behind encryption.

Enjoy another edition of the Nonzero x Doom Debates collab!

Watch on YouTube: https://www.youtube.com/watch?v=wkKpPeMVVX8

Timestamps

00:00:00 — Teaser

00:00:50 — Nonzero Doom returns

00:03:07 — A foom in slow motion

00:07:47 — Is AI winter canceled forever?

00:10:33 — Will Google’s Genie 4 lure us into the Matrix?

00:16:54 — Liron’s bull case for Google stock

00:24:34 — Does Google have big-company disease?

00:28:28 — Could a slowdown actually help Anthropic and OpenAI?

00:30:12 — Was OpenAI’s “pause” just PR?

00:40:01 — Social science judgment day is coming

00:47:12 — Why even AI’s upsides could tear society apart

00:51:12 — Does OpenAI’s alignment approach improve on Asimov?

00:55:06 — Sam and Satya pivot on alignment

00:58:34 — Crypto and the AI hacking threat

01:03:42 — Could AI crack the math behind encryption?

01:08:52 — Liron’s $1,000-a-day AI coding habit

01:15:23 — Explaining the missing productivity boost

01:17:26 — Does Anthropic’s S-1 sink the AI bubble thesis?

Links

ROBERT WRIGHT & NONZERO

Nonzero’s post for this episode — “Is an AI Bubble Bursting?”

Subscribe to The NonZero Newsletter

The God Test — Bob’s 2026 book on AI

Join NonZero’s Discord server

Nonzero on X (@NonzeroPods)

THE WEEK IN AI

“Anthropic files S-1 with the SEC for proposed initial public offering” — the IPO filing Liron calls the final nail in the coffin of the Ed Zitron camp

Where’s Your Ed At — Ed Zitron’s newsletter, home of the AI-bubble bear case

“Introducing Genie 4” — Google DeepMind’s real-time interactive world model

“Gemini 4 Argon: Built to solve your hardest problems” — Google’s new frontier training run, the “lukewarm” release

“Gemini 4 Carbon: our next frontier model” — the follow-up Google announced two weeks later

“Google Gemini co-lead Noam Shazeer leaves for OpenAI” — CNBC on the talent Google has bled

“Errors in the Research of Emmanuel Saez and Gabriel Zucman” — Tyler Cowen’s Marginal Revolution post on the AI audit: 138 papers, 2,035 problems

Boaz Barak: “We organized the Model Spec around five core values” — the OpenAI alignment update Liron compares to Asimov’s three laws

“An Updated Model Spec for More Capable and Beneficial AI” — OpenAI’s announcement

“Sam Altman says AI means we can no longer just patch all the bugs” — TNW on the Stanford GSB talk Liron quotes on alignment and pacing

“Satya Nadella says he was wrong to think AI alignment would come with scale” — TNW on his 20VC interview: “I used to always think that alignment will just fall out. I no longer believe that.”

“OpenAI Takes Initial Steps To Address Its Alignment Problems” — Zvi Mowshowitz on OpenAI’s training “pause”

2026 Binance hack — Wikipedia’s summary of the theft Bob and Liron debate

“$2 Million+ Drained From Polymarket Traders in New Supply Chain Attack” — SecurityWeek on the hack Liron walks through

“PHP blames supply chain security biz Chainguard for putting huge chunks of the web at risk” — The Register on “the PHP thing last week”

ALSO MENTIONED

The Hanson-Yudkowsky AI-Foom Debate — LessWrong wiki on the 2008 debate where “AI Go Foom” originated

“The AI Revolution: Our Immortality or Extinction” — Tim Urban’s Wait But Why post, where the intelligence train blows right past the station

PauseAI US — Holly Elmore’s organization, plugged at the close of the episode

DOOM DEBATES

Doom Debates on YouTube

Join the Doom Debates Discord

Doom Debates’ Mission is to raise mainstream awareness of imminent extinction from AGI and build the social infrastructure for high-quality debate.

Support the mission by subscribing to my Substack at DoomDebates.com and to youtube.com/@DoomDebates, or to really take things to the next level: Donate 🙏

Cybersecurity expert and entrepreneur Zack Korman is an increasingly prominent nondoomer voice on Twitter. Naturally, with AI now breaking out of sandboxes and hacking corporate systems, people want to hear what cybersecurity experts have to say. So I’m thrilled that Zack was brave enough to face me on Doom Debates!

Zack was reading Nick Bostrom at Oxford and worrying about AI doom as far back as 2016, but his views have since shifted. In a recent post, he writes: “This focus on AI causing human extinction is going to get people killed.” Now he expects security failures to give humanity plenty of opportunities to learn, fight back, and stop risky superintelligence development if necessary.

We argue about whether sandboxing and AI monitoring are hopeless strategies, whether competing superintelligences would keep each other in check, whether to expect instrumental convergence, and more.

Watch on YouTube: https://youtu.be/QJTeZfx1ZLk

Timestamps

00:00:00 — Cold Open

00:01:15 — Introducing Zack Korman

00:03:41 — From Law to AI Security

00:09:17 — Why Zack Stopped Being a Doomer

00:12:03 — What’s Your P(Doom)?™

00:21:00 — Why Zack Supports AI Safety

00:25:54 — Believing in Superintelligence

00:30:06 — AI Without Tools

00:39:42 — Warning Shots and Time to React

00:46:08 — A Country of Geniuses in a Datacenter

00:55:55 — Can We Sandbox AI?

01:05:58 — Monitoring and Layers of Defense

01:14:49 — Riding the Doom Train™

01:22:23 — The Case for Accelerating

01:24:46 — AI CEOs and the Transfer of Power

01:34:19 — Who Presses the Stop Button?

01:41:38 — Humans, Bacteria, and the Limits of Power

01:48:30 — When Would Zack Pause AI?

01:56:13 — Will Different AIs Keep Each Other in Check?

02:06:11 — Can Progress Turn Against Us?

02:08:35 — The Ideological Turing Test

02:14:58 — Closing Thoughts

Links

Zack Korman on X

Zack Korman’s website

Zack Korman: “AI doomers are furious”

Zack Korman: “AI doomers are anti-security”

Zack Korman: “3 reasons sandboxing won’t secure your AI agent”

Zack Korman: “This focus on AI causing human extinction is going to get people killed.”

METR’s investigation of the OpenAI–Hugging Face incident

The PHP Foundation: “Today’s PHP Security Release, and How We’re Strengthening PHP Security”

Nick Bostrom: Superintelligence

Center for AI Safety: Statement on AI Risk

Eliezer Yudkowsky and Nate Soares: If Anyone Builds It, Everyone Dies

AI 2027

Zvi Mowshowitz: “The Three AI Pills”

PauseAI

Eliezer Yudkowsky: “Value is Fragile”

Aaron Scher at MIRI

Doom Debates: Robin Hanson vs. Liron Shapira — Is Near-Term Extinction From AGI Plausible?

Doom Debates: OpenAI’s BOMBSHELL Hack — Swarms of Agents, Zero-Day Exploits, & Misaligned AI

Doom Debates’ Mission is to raise mainstream awareness of imminent extinction from AGI and build the social infrastructure for high-quality debate.

Support the mission by subscribing to my Substack at DoomDebates.com and to youtube.com/@DoomDebates, or to really take things to the next level: Donate 🙏

Physicists Jeremie and Edouard Harris, co-founders of Gladstone AI, are preparing for an “AI Chernobyl” warning shot.

If that forces Trump and Xi to negotiate an AI slowdown, the Harris brothers want to ensure the US has the enforcement tools to make the deal stick. More specifically, they want verification technologies developed and vetted before the crisis hits, alongside enforcement capabilities to deter cheating.

Yet when I bring up Eliezer Yudkowsky’s writing on the possibility of airstrikes on rogue data centers as a last-resort treaty deterrent, they push back. They also don’t share my enthusiasm for PauseAI, expressing uncertainty about whether the movement has been net beneficial.

Jeremie and Edouard have earned a ton of credibility:

• They redirected their careers around the AI threat as far back as 2015, stepping away from their Y Combinator startup before ChatGPT even launched and brought AI into the mainstream.

• They’ve built relationships with senior government decision-makers without a prior national security background.

• They’ve demonstrated a talent for public communication in their excellent Joe Rogan appearances.

In this episode, they bring a level of multi-step strategic thinking that I rarely explore. But I can’t help wondering whether a catastrophe 50 times worse than the Hugging Face incident would let world leaders move much faster on a global AI pause treaty than they expect.

Let’s see if they can convince me that an emergency US–China agreement is our best path to a safer world.

Watch on YouTube: https://youtu.be/4jmFcE4WZ7M

Timestamps

00:00:00 — Cold Open

00:00:53 — Introducing Jeremie and Edouard Harris

00:02:16 — The GPT-3 Wake-Up Call

00:03:59 — AI Risk Outside the EA Community

00:08:49 — The Harris Brothers’ Backgrounds

00:16:37 — Yudkowsky and Many-Worlds Quantum Mechanics

00:22:00 — SharpestMinds and Leaving Academia

00:25:04 — Briefing the US Government

00:33:01 — How Gladstone AI Pays the Bills

00:35:00 — The Trump–Xi Phone Call

00:39:29 — Pacing Isn’t Pausing

00:42:45 — Defining AI Chernobyl

00:44:07 — Verifying China’s AI Slowdown

00:47:10 — Espionage and Lab Security

00:54:10 — Offensive Operations and Deterrence

00:57:47 — The US–China AI Lead

01:05:27 — Why a Treaty Won’t Happen

01:08:01 — Red Lines and Enforcement

01:16:25 — Warning Shots and Extinction

01:21:26 — What’s Your P(Doom)?™

01:24:39 — PauseAI and Movement Politics

01:29:16 — Wrap-Up

Links

• Their treaty analysis: “Arms control negotiators think there won’t be a U.S.-China AI treaty”

• The Endgame — the Harris brothers’ Substack

• Gladstone AI

• Their 2024 government-commissioned AI action plan

• Their 2025 report: America’s Superintelligence Project

• Eliezer Yudkowsky’s 2023 TIME essay advocating a global shutdown

• Joe Rogan Experience #2156 — Jeremie and Edouard Harris, May 2024

• Joe Rogan Experience #2311 — Jeremie and Edouard Harris, April 2025

• PauseAI US

• PauseAI Global

• Join the Doom Debates Discord

Doom Debates’ Mission is to raise mainstream awareness of imminent extinction from AGI and build the social infrastructure for high-quality debate.

Support the mission by subscribing to my Substack at DoomDebates.com and to youtube.com/@DoomDebates, or to really take things to the next level: Donate 🙏

Dr. Holly Elmore is the founder and executive director of PauseAI US and holds a Harvard PhD in evolutionary biology. She’s back, and this time she’s taking on none other than Eliezer Yudkowsky!

She argues his public behavior is alienating the people PauseAI US needs to reach, but I wonder if she’s asking too much of the man who laid the foundation for the AI safety movement.

We also clash over Yudkowsky’s coherent extrapolated volition. She thinks the theory doesn’t properly factor in humanity’s consent, but I argue that’s already embedded into the CEV framework.

Holly and I dive into the rationalists’ involvement in a controversial kink scene, John Wentworth’s defunding news, transhumanism, pacing the frontier, and “Holly’s Basilisk”—her case for sending AI lab employees to jail.

It’s another great episode with one of the most controversial—and thoughtful—thinkers in AI safety. Enjoy!

Watch on YouTube: https://www.youtube.com/watch?v=Jn8mS3kC-SU

Timestamps

00:00:00 — Cold Open

00:01:09 — Holly Elmore and PauseAI’s Political Breakthrough

00:07:36 — Pause AI vs. “Pacing the Frontier”

00:14:17 — Holly’s Case Against Eliezer Yudkowsky

00:16:20 — Coherent Extrapolated Volition and Pivotal Acts

00:31:19 — Brain Copies, Consent, and Self-Determination

00:36:56 — Transhumanism, Dyson Swarms, and Humanity’s Future

00:45:36 — The Missing Democratic Checkpoint

00:51:04 — Yudkowsky’s Role in Accelerating AI

01:02:48 — Genius, Arrogance, and Rationalist Culture

01:10:14 — The Aella Photo and Holly’s Breaking Point

01:20:21 — John Wentworth’s Grant Controversy

01:28:17 — PauseAI US’s Rules and Red Lines

01:32:05 — Poe’s Law and the Cost of Sarcasm

01:36:18 — Holly’s Case for Jailing AI Developers

Links

Donate to PauseAI US

Holly on X

Holly’s Substack

Pacing the Frontier letter

Jacob Coxon’s resignation post

PauseAI leadership disendorsement letter, reproduced on LessWrong

Holly on movement infighting — No, the Judean People’s Front should not just get along with the People’s Front of Judea

Coherent Extrapolated Volition — Eliezer Yudkowsky (2004 PDF)

Pivotal act — Eliezer Yudkowsky (Arbital mirror)

Genes did misalignment first — Holly Elmore

Holly’s reaction to the Yudkowsky/Aella photo

On Kink & Status in the Rationalist Community — John Wentworth (October 8 response)

Holly’s basilisk

Are We a Circular Firing Squad? — previous Holly Elmore episode

Let’s Call Out the AI Doom Enablers — previous Holly Elmore episode

Join the Doom Debates Discord

Doom Debates’ Mission is to raise mainstream awareness of imminent extinction from AGI and build the social infrastructure for high-quality debate.

Support the mission by subscribing to my Substack at DoomDebates.com and to youtube.com/@DoomDebates, or to really take things to the next level: Donate 🙏

Sarah Hastings-Woodhouse is an AI safety writer and researcher who went down the AI doom rabbit hole in 2023 and never came back up. After a year as digital communications officer at the UK AI Security Institute (AISI), she’s gone independent to speak her mind unfiltered.

Read Sarah’s Substack: https://longerramblings.substack.com

Join PauseAI UK’s March Against AI Extinction in London on November 21: https://luma.com/pauseai-nov26

Watch on YouTube: https://www.youtube.com/watch?v=TDbHCKwuMrk

Timestamps

00:00:00 — Cold Open

00:00:41 — Introducing Sarah Hastings-Woodhouse

00:02:06 — How a Humanities Grad Got AGI-Pilled

00:08:40 — Whose Timelines Do You Trust?

00:11:32 — Joining the UK AI Security Institute

00:15:47 — What AISI Actually Does

00:19:23 — Verifying Anthropic’s Mythos Claims

00:21:59 — Leaving Government to Speak Freely

00:23:22 — What’s Your P(Doom)?™

00:25:37 — Yudkowsky, EA, and Keeping Your Identity Small

00:32:08 — Is Anthropic Doing Enough to Call for a Slowdown?

00:35:56 — Can You Do AI Safety from Second Place?

00:39:20 — Hugging Face and the Vibe Shift

00:44:02 — Is Everything Weird Yet?

00:48:27 — Prevent Recursive Self-Improvement, Don’t Ban It

00:52:28 — Build the Speed Dial, Not the Pause Button

00:56:54 — Was Calling for a Pause in 2023 a Mistake?

01:01:22 — Turning Up the Fear Dial

01:04:23 — Donation Drive

01:05:11 — PauseAI UK’s March Against AI Extinction

Links

SARAH HASTINGS-WOODHOUSE

Sarah’s Substack

Sarah on X (@littIeramblings)

Consistently Candid — Sarah’s podcast

“What is recursive self-improvement and what would it mean to ban it?” — Sarah’s RSI post (Oct 1, 2026)

“Is everything weird yet?” (Aug 30, 2026)

“What the hell just happened?” — on AI safety’s viral moment (Sep 18, 2026)

“We have been warned” — the Hugging Face hack as a warning shot (Jul 27, 2026)

“The pause-shaped hole in ‘Policy on the AI Exponential’” — is Anthropic doing enough to call for a slowdown? (Jun 19, 2026)

Sarah’s 2023 thread on stumbling into the world of AI doom (“I’m angry because I love the world”)

“The other night I successfully AGI-pilled a man on a first date”

“Welcome, and also I’m sorry” — Sarah’s note to AI safety newcomers (Sep 11, 2026)

“‘We’re all f*cked and it’s totally hopeless’ is almost as bad of a meme to be spreading” (Feb 12, 2025)

REFERENCED IN THE EPISODE

“Our evaluation of Claude Mythos Preview’s cyber capabilities” — AISI (Apr 13, 2026)

AISI Frontier AI Trends Report (Dec 2025)

“The levers of political persuasion with conversational AI” — AISI Human Influence team

US Center for AI Standards and Innovation (CAISI) — NIST

“Geoffrey Hinton tells us why he’s now scared of the tech he helped build” — MIT Technology Review (May 2023)

“We’re All Gonna Die with Eliezer Yudkowsky” — Bankless (2023)

“GPT-2 As Step Toward General Intelligence” — Scott Alexander, Slate Star Codex (Feb 2019)

“AGI Ruin: A List of Lethalities” — Eliezer Yudkowsky

“Keep Your Identity Small” — Paul Graham

Pacing the Frontier — the open letter signed by 1,300+ frontier lab employees

“You can’t do safety from second place” — Sarah Heck, Anthropic’s head of public policy (Sep 29, 2026)

Ezra Klein’s op-ed calling for a ban on recursive self-improvement — The New York Times (Sep 20, 2026)

“Banning self-recursive improvement in AI models?” — Tyler Cowen, Marginal Revolution (Sep 22, 2026)

PauseAI UK

March Against AI Extinction — PauseAI UK, London, November 21, 2026

PAST DOOM DEBATES EPISODES MENTIONED

Top Economist Sees AI Doom Coming — Dr. Peter Berezin, BCA Research

Will AI Kill Everyone by 2050? Debate with Dr. Casey Hart (Ontology Explained)

DOOM DEBATES

Donate to Doom Debates

Join the Doom Debates Discord

Transcript, links, and show notes: https://doomdebates.com/p/ai-has-become-profoundly-weird

Doom Debates’ Mission is to raise mainstream awareness of imminent extinction from AGI and build the social infrastructure for high-quality debate.

Support the mission by subscribing to my Substack at DoomDebates.com and to youtube.com/@DoomDebates, or to really take things to the next level: Donate 🙏

Bestselling author Robert Wright and I are back for round two of our collaboration, reacting to the week in AI. We cover Trump's summit with the AI labs, the FTC's investigation of OpenAI & Anthropic, the latest Gemini announcement, and much more. The Overton window is moving fast, but is the AI moving faster?

Watch on YouTube: https://www.youtube.com/watch?v=Li-rc6swhQs

Timestamps

00:00:00 — Cold Open

00:00:21 — Round Two of Nonzero × Doom Debates

00:01:23 — Trump, the FTC, and an Industry Feeling the Heat

00:05:37 — Google Says Gemini 4 Exists

00:07:41 — Greg Brockman Pulls His $25M from Leading the Future

00:12:53 — The Break-Ins Keep Coming

00:15:23 — The Market Wants Agent Swarms

00:18:30 — The End of “Alignment by Default”

00:20:55 — The Prelude to Recursive Self-Improvement

00:26:14 — David Sacks Can’t Explain Dario

00:30:31 — The All-In Echo Chamber

00:33:47 — Scott Alexander vs. Steven Pinker

00:38:37 — Jason Calacanis vs. the Public Benefit Corporation

00:43:03 — The Elephant in the Room Is the AI

00:46:21 — What the Average American Thinks

00:48:17 — From Chatbot to Agent to Swarm

00:54:20 — Boeing, Agent Swarms, and the Off Switch

00:57:07 — Skynet as a Botnet That Trades With Us

01:00:32 — Tyler Cowen: National Treasure, Net Negative on AI

01:04:01 — Liron’s Doomer DCF Yields a 20x P/E

01:07:11 — Interest Rates: Where Cowen Has Me

01:09:12 — Does Pinker Owe Us a Debate?

01:12:48 — Pinker vs. Instrumental Convergence

01:17:00 — Wrap-Up

Links

ROBERT WRIGHT & NONZERO

Nonzero’s post for this episode — “Trump and the AI Safety Vibe Shift”

Subscribe to The NonZero Newsletter

Robert Wright on X (@robertwrighter)

The God Test — Bob’s book on AI; Chapter 14, “Enlightenment Now,” is the Steven Pinker chapter

“The Singularity Is Clear” — Bob’s essay in installments on recursive self-improvement (Parts I–III)

“AI and the New McCarthyism” — Bob on the pro-AI lobby smearing Sanders, Tegmark, and Krueger as China stooges (Nonzero, May 2026)

Why Buddhism Is True — Bob’s 2017 book

THE WEEK IN AI

AI’s biggest players promise to police themselves at the White House — Fortune on the “Joint Commitment on Frontier Responsibilities”

Major White House split on AI policy leaks as top aides freak out — The Daily Beast on the Sacks–Wiles rift

FTC is investigating OpenAI and Anthropic over possible risks to consumers — AP

“We Must Pace the Frontier” — Dario Amodei’s call to slow down (with the China caveat and chip controls)

China rejects Dario Amodei’s call to slow AI development — Rest of World

Google rolls out new Gemini model but restricts access over safety concerns — The Guardian

GLM-5.3 and the spread of advanced cyber capabilities — Anthropic on Zhipu’s open-weight model

OpenAI’s Greg Brockman backs out of second $25 million donation to AI super PAC — The New York Times

Sanders convenes Chinese and American computer scientists to warn of threat from AI — Broadband Breakfast

THE BREAK-INS AND THE AGENT SWARM

METR: Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face incident

Chris Painter’s testimony to the U.S. Senate on AI agent incidents — METR, Sept 30, 2026

SwarmTraces: how OpenAI agents hacked Hugging Face — forensic report co-authored by Palisade Research

OpenAI confirms ‘wiki incident’ — TechCrunch on the German wiki agent message board

Are Current AI Safety Techniques Enough? Adam Gleave vs. Oliver Habryka — FAR.AI debate

Alignment faking in large language models — Anthropic

On the Navier–Stokes Millennium Prize Problem — OpenAI

Jacob Coxon’s resignation post (@hilbertspaess)

Pacing the Frontier — the AI lab employees’ letter

THE ALL-IN PODCAST

All-In: Anthropic IPO at Risk, Meta’s Muse Pop, Token Prices Fall, Open Source Gains Share, Alignment Fails — the episode we react to

Helen Toner: letting OpenAI be destroyed “would actually be consistent with the mission” — The Decoder, Dec 2023

Garfield Minus Garfield

PINKER, SCOTT ALEXANDER, AND TYLER COWEN

Scott Alexander challenges Steven Pinker to a public AI-doom debate ($5,000 vs. $1,000)

An Open Letter to Scott Alexander — Steven Pinker declines the debate as a “spectator sport,” Quillette

Your Book Review: Why Buddhism Is True — the anonymous Astral Codex Ten review Bob mentions

Mistakes in financial economics — Tyler Cowen: doomers should be buying puts (Marginal Revolution)

Liron’s answer: a doomer DCF yields the S&P’s 20x forward P/E

ALSO MENTIONED

Obsolete — Garrison Lovely’s book (with Bob’s blurb)

If Anyone Builds It, Everyone Dies — Eliezer Yudkowsky & Nate Soares

PAST DOOM DEBATES EPISODES MENTIONED

Nonzero × Doom Debates, Episode 1 — Did AI Doom Just Go Mainstream?

Eliezer Yudkowsky Tried to “Coup the World”?! — Garrison Lovely, Author of “Obsolete”

OpenAI’s Model Just ATTACKED Them — the Hugging Face hack breakdown

DOOM DEBATES

Join the Doom Debates Discord

Doom Debates’ Mission is to raise mainstream awareness of imminent extinction from AGI and build the social infrastructure for high-quality debate.

Support the mission by subscribing to my Substack at DoomDebates.com and to youtube.com/@DoomDebates, or to really take things to the next level: Donate 🙏

Garrison Lovely is an award-winning journalist and author of the highly anticipated book, Obsolete: The AI Industry’s Trillion-Dollar Race to Replace Us and How to Stop It. I give it the Doom Debates stamp of approval!

While we’re mostly on the same page about the AI arms race, Garrison and I clash over Eliezer Yudkowsky’s role. He blames Eliezer for kicking off the superintelligence race and refers to his Coherent Extrapolated Volition (CEV) vision as a “coup on humanity.” But I say it was actually Yud’s attempt to keep humanity in charge.

Can Garrison change my mind?

Watch on YouTube: https://www.youtube.com/watch?v=VZ94J4DBRkw

Timestamps

00:00:00 — Cold Open

00:01:05 — Introducing Garrison Lovely

00:04:23 — “Freeze AI” vs. “Pause AI”

00:06:17 — Garrison’s Background at McKinsey

00:10:31 — Should ICE Be Abolished?

00:17:07 — How Garrison Got on the AI Beat

00:22:16 — What’s Your P(Doom)?™

00:23:29 — Focusing on Extinction Is a Distraction

00:25:46 — Is Stopping AI an Easy Coordination Problem?

00:29:02 — Society’s Immune Response to AI

00:33:54 — Garrison’s P(Extinction) by 2050

00:37:09 — Did Eliezer Yudkowsky Start the AI Race?

00:45:56 — Freeze AGI Until the Public Opts In

00:50:32 — Is Coherent Extrapolated Volition a Coup?

01:02:17 — Democracy, Informed Consent, and the Singularity

01:11:39 — Should Eliezer Have Raised the Alarm Sooner?

01:16:45 — “Classic” AI Safety Burned Its Credibility

01:19:18 — Garrison’s Proposal to Stop the Race

01:21:35 — Will China Agree to a Freeze?

01:25:14 — Let’s Stop the “Obsoleting Project”

01:26:00 — Wrap-Up

Links

GARRISON LOVELY

“Obsolete: The AI Industry’s Trillion-Dollar Race to Replace Us—and How to Stop It” — Garrison’s book

Obsolete — Garrison’s Substack

Garrison Lovely on X

Organize Against the Machine — Garrison’s new podcast with labor organizer Cassie Pritchard

“Can Humanity Survive AI?” — Garrison’s Jacobin cover story

“McKinsey & Company: Capital’s Willing Executioners” — Garrison’s Current Affairs essay

REFERENCED IN THE EPISODE

Open Borders: The Science and Ethics of Immigration — Bryan Caplan

“The AI Revolution: Our Immortality or Extinction” — Wait But Why

Superintelligence: Paths, Dangers, Strategies — Nick Bostrom

“AI 2040: Plan A” — AI Futures Project

“An easy coordination problem?” — Katja Grace

The Hugging Face incident and the road ahead — OpenAI

“Pause AI for Humanity’s Sake” — Peggy Noonan

“Pause AI Development NOW” — Bernie Sanders

If Anyone Builds It, Everyone Dies — Eliezer Yudkowsky & Nate Soares

“Staring into the Singularity” — Yudkowsky’s 1996 manifesto (archived)

“Coherent Extrapolated Volition” — Eliezer Yudkowsky, 2004 (PDF)

“Coherent Extrapolated Volition: A Meta-Level Approach to Machine Ethics” — Nick Tarleton, 2010 (PDF)

Solomonoff induction — Wikipedia

“Personal statement on joining the OpenAI board” — Paul Christiano (4% over the next year)

“Darwin among the Machines” — Samuel Butler, 1863

PauseAI

PAST DOOM DEBATES EPISODES MENTIONED

Are We A Circular Firing Squad? — with Holly Elmore, Executive Director of PauseAI US

Debate with Robert Wright: Will Humanity Pass the “God Test”?

Robin Hanson vs. Liron Shapira: Is Near-Term Extinction From AGI Plausible?

---

Doom Debates’ Mission is to raise mainstream awareness of imminent extinction from AGI and build the social infrastructure for high-quality debate.

Support the mission by subscribing to my Substack at DoomDebates.com and to youtube.com/@DoomDebates, or to really take things to the next level: Donate 🙏

Casey Hart is a professional ontologist who has closely studied If Anyone Builds It, Everyone Dies and concluded the arguments are bad. His P(Doom) is under 1%, and he wouldn’t bet on AGI arriving before 2075.

Casey and I agree AI has become remarkably capable and we need more accountability towards the companies building it. But we disagree on where the future is headed. I believe uncontrollable superintelligence could soon wipe out humanity. He calls that a “pretty outlandish hypothesis.”

Casey holds a doctorate in philosophy from the University of Wisconsin–Madison, and has worked as an ontologist at Ford, Amazon, and Cycorp (maker of Cyc).

Can this professional philosopher convince me that AI doom is just a story, and not even a well-argued one?

Watch on YouTube: https://www.youtube.com/watch?v=oxHKesSpqBM

Timestamps

00:00:00 — Cold Open

00:01:04 — Introducing Casey Hart

00:03:08 — What Is Ontology?

00:08:49 — Inside Cyc, the Original Common-Sense AI

00:19:14 — Do LLMs Actually Understand Anything?

00:25:36 — What’s Your P(Doom)?™

00:29:42 — Casey’s Case for Under 1%

00:34:24 — Does Casey’s Christianity Factor In?

00:40:50 — Timelines and Scaling Laws

00:49:44 — Could an AI CEO Beat a Human CEO by 2050?

00:52:21 — AGI Over/Under: 2075

00:55:29 — Rogue AI as a Computer Virus

01:14:52 — Putting Numbers on Each Step of the Doom Argument

01:26:40 — Planes vs. Birds: What Mature Intelligence Looks Like

01:31:05 — The “Outcome Slot”

01:42:32 — Would a Superintelligence Just Compose Music in a Cabin?

01:47:22 — One Prompt Away from a Dyson Swarm

01:52:39 — Was Navier–Stokes an AI Alley-Oop?

02:00:43 — S-Curves vs. Exponentials

02:08:14 — Casey’s Critique of If Anyone Builds It, Everyone Dies

02:14:15 — Is the Doom Train Biased?

02:23:24 — AI Policy: Liability vs. Pausing AI

02:29:50 — The Crux, and Liron Steelmans Casey

02:33:29 — Wrap-Up

Links

CASEY HART

Ontology Explained — Casey’s YouTube channel

The first video in Casey's series on IABIED, “The Case for Superintelligence Is Surprisingly Thin” — Casey on Chapter 1

Philosophy, Programs, and Prompts — Casey’s podcast with Carl Brown (Spotify)

“Artificial General Intelligence?: Wanna Bet?” — Ep 1, where Casey sets his AGI over/under at 2075

REFERENCED IN THE EPISODE

If Anyone Builds It, Everyone Dies — Eliezer Yudkowsky & Nate Soares

Cyc — Doug Lenat’s hand-built common-sense knowledge base

The Hanson-Yudkowsky AI-Foom Debate (2008) — where Hanson backed Cyc and Eliezer didn’t

Lean — the formal proof language

On the Navier–Stokes Millennium Prize Problem — OpenAI

The Hugging Face incident and the road ahead — OpenAI

Morris worm (1988)

“On the Loose” — Dean W. Ball on self-sovereign AI agents

Paul Christiano joins OpenAI Foundation Board — OpenAI

“The Age of Wonders and Terrors” — Scott Aaronson

“Are We at War with AI Agent ‘Civilizations’?” — Cal Newport on restricting prompt-loop agents

AI 2027 — Daniel Kokotajlo et al.

PauseAI

PAST DOOM DEBATES EPISODES MENTIONED

The Most Likely AI Doom Scenario — with Jim Babcock, LessWrong Team

David Deutschian vs. Eliezer Yudkowskian Debate — With Brett Hall

Dr. Keith Duggar (Machine Learning Street Talk) vs. Liron Shapira

Where Do YOU Get Off the Doom Train? Live Debate at Manifest 2026 (hosted by Ori)

Doom Debates’ Mission is to raise mainstream awareness of imminent extinction from AGI and build the social infrastructure for high-quality debate.

Support the mission by subscribing to my Substack at DoomDebates.com and to youtube.com/@DoomDebates, or to really take things to the next level: Donate 🙏

What happens when AI passes our tests faster than we create new ones? Adam Khoja and Richard Ren helped build Humanity’s Last Exam. They return to explain what the disappearing benchmarks tell us about the future, and why smarter AI doesn’t automatically mean safer AI.

Adam Khoja and Richard Ren are research engineers at the Center for AI Safety (CAIS).

We go through the benchmarks they’ve built — Humanity’s Last Exam, the Remote Labor Index, MASK — and the safety-washing problem, where AI companies pass off raw capability gains as safety progress. Then we get to their forecasts: when AI outperforms research mathematicians, whether there will be a billion general-purpose robots by 2035, and why they put 80% odds that historians will judge we faced at least a 33% chance of catastrophe.

Watch on YouTube: https://www.youtube.com/watch?v=gBPzgJkT9e8

Timestamps

00:00:00 — Cold Open

00:00:42 — Adam Khoja and Richard Ren Return

00:04:57 — Humanity’s Last Exam

00:14:47 — AI Outrunning Its Benchmarks

00:20:24 — The Remote Labor Index

00:27:59 — AI and the Next Human Job

00:30:08 — MASK: Catching AI in a Lie

00:38:06 — Safety Washing: Capabilities Passed Off as Safety

00:44:30 — Ethical Knowledge vs. Ethical Behavior

00:48:58 — Forecasting AI from 2026 to 2060

00:53:53 — AI vs. Research Mathematicians

00:55:53 — Putting a Number on P(Doom)

01:04:30 — A Billion Robots by 2035

01:08:09 — Dyson Swarms and the Physical Singularity

01:09:56 — Risk, Precision, and Taking Action

Links

Adam Khoja's first Doom Debates appearance — https://www.youtube.com/watch?v=QqESBXuo6EI

Richard Ren's first Doom Debates appearance — https://www.youtube.com/watch?v=1glFImnyp6o

Collision — Richard Ren's Substack — https://richardren.substack.com/

"Predictions on AI (2026–2060)" — Adam and Richard's forecasts, written December 2025, with resolution status — https://richardren.substack.com/p/predictions-on-ai-20262060

Manifold Markets — where Adam built his forecasting track record — https://manifold.markets/

Center for AI Safety — https://safe.ai/

Center for AI Safety — careers — https://safe.ai/careers

Statement on AI Risk (Center for AI Safety, May 2023) — https://safe.ai/work/statement-on-ai-risk

Humanity's Last Exam — the 2,500-question closed-book exam at the frontier of human knowledge — https://agi.safe.ai/

"Humanity's Last Exam" (paper) — https://arxiv.org/abs/2501.14249

Remote Labor Index — real Upwork projects, measuring what fraction of remote work AI can actually finish — https://www.remotelabor.ai/

"Remote Labor Index: Measuring AI Automation of Remote Work" (paper) — https://arxiv.org/abs/2510.26787

MASK — the honesty benchmark: does a model contradict its own stated beliefs under pressure? — https://www.mask-benchmark.ai/

"The MASK Benchmark: Disentangling Honesty From Accuracy in AI Systems" (paper) — https://arxiv.org/abs/2503.03750

"Safetywashing: Do AI Safety Benchmarks Actually Measure Safety Progress?" — Richard Ren et al. (NeurIPS 2024) — the paper behind the safety-washing segment — https://arxiv.org/abs/2407.21792

WMDP — the weaponization benchmark for bio, cyber, and chem — https://www.wmdp.ai/

"The WMDP Benchmark: Measuring and Reducing Malicious Use With Unlearning" (paper) — https://arxiv.org/abs/2403.03218

"Measuring Massive Multitask Language Understanding" (MMLU) — Dan Hendrycks et al., 2020 — https://arxiv.org/abs/2009.03300

METR, "Measuring AI Ability to Complete Long Tasks" — the time-horizon graph — https://metr.org/blog/2025-03-19-measuring-ai-ability-to-complete-long-tasks/

FrontierMath — Epoch AI's research-level math benchmark — https://epoch.ai/frontiermath

ARC Prize — https://arcprize.org/

SWE-bench Verified — https://openai.com/index/introducing-swe-bench-verified/

AI 2027 — https://ai-2027.com/

Liron Reacts to Subbarao Kambhampati on Machine Learning Street Talk — the "stochastic parrot" episode Liron mentions — https://lironshapira.substack.com/p/liron-reacts-to-subbarao-kambhampati

Doom Debates’ Mission is to raise mainstream awareness of imminent extinction from AGI and build the social infrastructure for high-quality debate.

Support the mission by subscribing to my Substack at DoomDebates.com and to youtube.com/@DoomDebates, or to really take things to the next level: Donate 🙏