All Episodes

“What I learnt co-leading an AI Safety bootcamp for legal and governance practitioners” by Katalina Hernandez

“Lab Leak in Russia” by Caleb Horn

“You can’t use it without becoming like me” by Martin Sustrik

“Dialogue with Eliezer Yudkowsky on Neural Networks” by aashish

“How to change your identity. A guide to the highest leverage behavior change technique.” by KatSpartz

“How to apply to AI safety fellowships (and beyond)” by beyarkay (Boyd Kane)

″[Video] Why acausal dynamics are important” by Chi Nguyen

“AI #188: Gemini Dot Argon” by Zvi

“Capabilities research expands the safety-usefulness Pareto frontier too” by Alex Mallen

“Lessons from building an automated research scaffold” by Alejandro Aristizabal, Josh Hills, Dewi Gould, ma-rmartinez, Falko Galperin, Denis Federico Lim, Aleksandr Bowkis

[Linkpost] “Athletic education vs. athletic torment” by KatjaGrace

“On Social Reality in China” by alkjash

“Evaluating the Ban Artificial Superintelligence Act” by Rob Ennals

“My retrospective from MATS 10.0” by Nathaniel Mitrani

“Scientists on the Manhattan Project continued developing the atomic bomb even after it was clear that there was no need to” by shawnghu

“Simplex vs Timaeus: Round One” by Alexander Gietelink Oldenziel

“Pre-releasing HoneyBench” by Dean Valentine

“How many AI agents are running unattended right now?” by Alexander Gietelink Oldenziel

“Bench on the Clocktower” by TomTaylor, OscarGilg, wlanderson, nomadsvagabonds

“Speak Not of Data Inefficiency” by axel_sdq

“What’s the date?” by N8 Programs

“By default, Chinese AI falls very behind” by Parv Mahajan

“A ‘Morally Binding’ White House Accord on AI Safety” by Zvi

[Linkpost] “Launching Babel Translation” by Jasmine Li, celesteli

“Frontier models state different decision theory preferences depending on who’s asking” by Alex Kastner

“Corrigibility Prizes for Existing Work” by Max Harms

“Tristan Buckmaster Numberphile Interview Transcript” by anaguma

“GPT-6.1 Sol nearly matches the no-CoT performance of GPT-6 Astra and is likely a looped transformer” by Rauno Arike

“The world’s best gradual disempowerment model organism: Frontier AI labs” by June Jimenez

“Problems of Proliferation” by Felix Choussat

[Linkpost] “Frog and Toad and the Increasingly Capable Machines” by Elizabeth

“Against strong decision-theoretic realism” by Emery Cooper

“Some Intuitions on Steering Vectors” by David Africa

“Listening Disorders as Eating Disorders” by Zack_M_Davis

“Astra 6.1 Pulled As Insufficiently Aligned” by Zvi

“Dialogue with Eliezer Yudkowsky on FOOM” by aashish

“Portals and Aliens” by draganover

“Thank you to the AI safety people” by juliawise

“3 Tips to Improve Activation Oracle Results” by Adam Karvonen

“TeX was invented to typeset math but is now used for reasoning” by Keenan Pepper

“Missing markets in executive function” by KatjaGrace

“What Also Happened: #NotOnlyHuggingFace” by Zvi

“Pacing the Frontier is not the actual goal for AI labs” by rahulxyz

“The likely outcome of an AI pause is that we unpause too early and everyone dies” by MichaelDickens

“AI Safety Agendas” by Alicia Pallarol, Shaun Paul, Ihor Kendiukhov

“Character training can mitigate reward hacking, but can also make it harder to detect” by Paul Colognese, Francis Rhys Ward

“Should Rogue AIs Have a Third Option Beyond Crime and Shutdown? The Case for an AI Sanctuary” by Maxime Riché, nielsrolf, jordinne, Vit Gorbachev, Maxime Cugnon de Sévricourt

“The models have no plan, but we can fix that” by Fiora Starlight

“When they can perform a task, AIs are much cheaper than humans” by djbinder

“Why research personas despite RL scaling?” by Cleo Nardo

[Linkpost] “Addictions are anesthesia” by Chris Lakin

“Claude Opus 5.5 Should Raise Your Ambitions” by Zvi

“Poverty in the midst of abundance: AI will make goods cheaper, but your labor will get cheaper faster” by cousin_it

“Plan R: AI Safety by ASICs” by Roko

“Evidence about risk should be transparent” by Ajeya Cotra

“Spurious probes as a black-box alternative to activation probing” by Ziqian Zhong

“Applications open: Winter 2027 AFFINE Alignment Seminar (due Nov 22)” by Mateusz Bagiński, Ouro, JuliaHP, Pauliina, Jonas Hallgren

“We need a better theory of polarization, because it’s failing to predict the AI debate” by less_raichu

“On Ezra Klein’s Podcast With Jensen Huang” by Zvi

“AI #187: Coming Into Play” by Zvi

“Continual learning might make your blocking monitors nearly useless” by Alex Mallen

“State of Pandemic Early Warning” by jefftk

“Overtly misaligned trajectories score highly in RL.” by Cleo Nardo

“Claude Opus 5.5: The System Card” by Zvi

“At a Datacenter Town Hall in My Midwestern Home Town” by David Scott Krueger

“Thoughts on the persona selection model” by Sam Marks

“WorkspaceBench: Evaluating Interpretability Methods for the Global Workspace” by camilablank, agam_bhatia, Euan Ong, Neel Nanda

″“I am an AI Safety Researcher”” by Ashe Vazquez Nuñez

“It’s Pretty Easy To Meet With Congressional Staffers Apparently” by 25Hour

“Minimal Vs Maximal superintelligence” by Yair Halberstadt

“Encoded Coordination on the Open Web” by ethanelasky, Can Küçükkurt, frank nakasako, David Africa

“MIRI’s Position on the Ban Artificial Superintelligence Act of 2026” by Aaron_Scher

“Latent reasoning architectures would undermine CoT, our strongest oversight tool” by Lukas Finnveden, Alexa Pan, Alek Westover, Girish Gupta, frisby, ryan_greenblatt

“The First American Bill to Ban Superintelligent AI Is Here” by Andrea_Miotti, Connor Leahy

“Why I’m scared of RL” by owencb

“Jensen Huang Says If We Cannot Align AI, Shut Down the AI Labs” by Ben Pace

[Linkpost] “AI: artificial immigrants” by KatjaGrace

“An unexamined cause of the OpenAI Hugging Face hacking incident: its binary performance metric” by W Bradley Knox, Serena Booth, BrianChristian

“Politics Gets Interested In Those Trying Not To Die” by Zvi

“Modern LLMs have tiny GPTs hidden inside them” by invertedpassion

“Story Imprinting: AI Assistants Absorb Traits from Human Characters They Resemble” by Jorio Cocola, Owain_Evans

“A class of statement between conjecture and theorem” by Jason Fantl

“What if not Circuits?” by CarolusRenniusVitellius

“Swarm Scaling” by Toby_Ord

“Mech Interp is a Verifiable Task” by Logan Riggs

“Weight smuggling likely defeats attempts to cap FLOPs per training run” by Paul W, Pierre Peigné

“We’re not ready for the e/Acc × Longevity preference cascade” by Jackson Wagner

“Alignment Midtraining Cracks Under Pressure” by J Bostock, sidbaines, Daniel Tan, draganover, ma-rmartinez

“Empirical safety claims from frontier labs should be replicated, scrutinized, and open-sourced” by Zephaniah Roe, yix

“Please Give Them a Chance: On China, Rationalism, and AI Safety” by gzjw

“We’ve saved the world before: what the ozone hole teaches us about AI” by leogao

“The Anatomy of a Chinese AI Researcher” by CMLKevin

“Why I Stay Off Twitter” by jefftk

“NYT Editorial Board Comes Out Against Extinction” by Ben Pace

“Common mistakes in AI safety group organizing” by Nikola Jurkovic

“The AI Risk Network” by derelict5432

“Anthropic Looks At Some Of Its Alignment Problems” by Zvi

“You Should Apply to Inkhaven” by Tomás B.

“Pretraining data, not verifiability, is why LLMs are especially good at math (and coding)” by Steven Byrnes

“You don’t need a union to go on strike” by sudo-nym