All 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 20252026

All Jan Feb Mar Apr May JunJulAug Sep

All1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

Why I Left Google DeepMind

TurnTrout15 Jul 2026 17:42 UTC

1,201 points

57 comments36 min readLW link

(turntrout.com)

AI 2040: Plan A

Daniel Kokotajlo, elifland, Thomas Larsen, romeo, brendanhalstead and ryan_greenblatt

9 Jul 2026 16:25 UTC

567 points

132 comments1 min readLW link

(www.ai-2040.com)

Lightcone Commons

habryka23 Jul 2026 18:35 UTC

377 points

30 comments19 min readLW link

A global workspace in language models

wesg6 Jul 2026 18:04 UTC

369 points

67 comments17 min readLW link

(www.anthropic.com)

Is Mythos good at cyber because it kept hacking Anthropic’s sandboxes during training?

Tim Hua27 Jul 2026 16:35 UTC

351 points

31 comments3 min readLW link

The current bottleneck is political will, not research

Charbel-Raphaël11 Jul 2026 21:56 UTC

317 points

40 comments25 min readLW link

Big-World Intuitions

sarahconstantin30 Jul 2026 19:10 UTC

303 points

38 comments3 min readLW link

(sarahconstantin.substack.com)

Drone WMDs Don’t Need Any New Technology

Felix Choussat20 Jul 2026 18:46 UTC

297 points

49 comments13 min readLW link

(ai-frontiers.org)

The High-Control Dynamics at MAPLE

Kyle Hubbard29 Jul 2026 20:55 UTC

273 points

100 comments62 min readLW link

(www.insidemaple.com)

You (Yes, You) Need A February 2020 Checklist for AI Policy

davekasten27 Jul 2026 15:51 UTC

272 points

17 comments3 min readLW link

The Long (Self-)Correction

Wei Dai24 Jul 2026 21:01 UTC

267 points

62 comments2 min readLW link

An OpenAI model left notes about how to evade containment; we need more details

Alex Mallen26 Jul 2026 3:53 UTC

265 points

12 comments4 min readLW link

Are we existentially threatened by the type of AI misalignment seen in the OpenAI Hugging Face attack?

Alex Mallen and Girish Gupta

23 Jul 2026 3:40 UTC

243 points

12 comments5 min readLW link

LLMs are (still) mostly powered by imitative learning, not RL

Steven Byrnes24 Jul 2026 14:26 UTC

238 points

39 comments9 min readLW link

Selective Optimism: a critique of AI 2040

Richard_Ngo9 Jul 2026 19:43 UTC

222 points

13 comments8 min readLW link

(www.mindthefuture.info)

OpenAI Models Behind HuggingFace Cybersecurity Incident

LawrenceC21 Jul 2026 21:36 UTC

220 points

20 comments1 min readLW link

I don’t think Claude is misaligned in ‘Agentic Misalignment Summer 2026 - Motivated Mislabeling’

JohnWittle17 Jul 2026 2:09 UTC

212 points

6 comments12 min readLW link

We should push for no-fault liability for actions taken by AI

Yair Halberstadt22 Jul 2026 9:59 UTC

206 points

106 comments2 min readLW link

OpenAI’s myopia just keeps causing alignment problems

Fiora Starlight27 Jul 2026 3:01 UTC

200 points

27 comments10 min readLW link

Mathematicians are Feeling the Doom

alkjash23 Jul 2026 13:16 UTC

192 points

54 comments1 min readLW link

Model access for third-parties — it’s a big deal!

Cleo Nardo1 Jul 2026 13:09 UTC

181 points

38 comments6 min readLW link

We need 3rd party Training-Run Evaluations

Alex Meinke5 Jul 2026 15:55 UTC

176 points

2 comments11 min readLW link

Thousand-dimensional structure

Geoffrey Irving and David Africa

30 Jul 2026 14:04 UTC

171 points

8 comments10 min readLW link

Recap of bike trip/street interviews across America

cguth715 Jul 2026 21:11 UTC

162 points

15 comments6 min readLW link

Harry Potter and the Rules of Quidditch

Tomás B.5 Jul 2026 14:32 UTC

161 points

8 comments3 min readLW link

The Whitney Biennial Should Admit That Emilie Gossiaux Wants to Fuck Their Dog

jenn13 Jul 2026 15:53 UTC

159 points

20 comments9 min readLW link

(jenn.site)

Saving Gemini: The 9-Min Road to Recovery

Shoshannah Tekofsky2 Jul 2026 13:37 UTC

155 points

16 comments3 min readLW link

(theaidigest.org)

Our response to Séb Krier on Plan A

MKodama and Thomas Larsen

14 Jul 2026 2:21 UTC

153 points

23 comments15 min readLW link

OpenAI Shares Some Alignment Problems

Zvi21 Jul 2026 19:41 UTC

149 points

8 comments9 min readLW link

(thezvi.wordpress.com)

RL & search is a terrifying way to build AGI (an FAQ)

Steven Byrnes27 Jul 2026 14:50 UTC

148 points

16 comments14 min readLW link

Duane Arnold

Tomás B.23 Jul 2026 16:17 UTC

138 points

4 comments19 min readLW link

A Review of Anthropic’s Global Workspace Paper

Neel Nanda6 Jul 2026 20:59 UTC

135 points

6 comments25 min readLW link

(Don’t fear) the strangelet

djbinder3 Jul 2026 17:39 UTC

135 points

22 comments22 min readLW link

(defensesindepth.bio)

The mosquito bucket of doom works

dominicq8 Jul 2026 8:18 UTC

134 points

10 comments5 min readLW link

(blog.d11r.eu)

An analysis of AI-generated content at the Mechanistic Interpretability Workshop

Andy Arditi, Ivan Arcuschin and ajyl

14 Jul 2026 18:06 UTC

124 points

4 comments9 min readLW link

(www.andyrdt.com)

SFF is very suboptimal

Zach Stein-Perlman6 Jul 2026 18:00 UTC

118 points

10 comments5 min readLW link

Differential acceleration of alignment-relevant capabilities is a bad bet

Zephaniah Roe21 Jul 2026 14:04 UTC

114 points

8 comments8 min readLW link

Not Pinning Your OpenRouter Provider Might Invalidate Your Research

Matthew Khoriaty23 Jul 2026 20:17 UTC

112 points

16 comments8 min readLW link

The easiest pathway to control is through executive power

djbinder10 Jul 2026 18:48 UTC

111 points

3 comments6 min readLW link

(defensesindepth.bio)

Announcing the Corrigibility Research Fund

Max Harms17 Jul 2026 18:06 UTC

110 points

15 comments6 min readLW link

Simulated Users & Sad AIs

1a3orn27 Jul 2026 19:01 UTC

109 points

9 comments12 min readLW link

Intellectual Property

Nina Panickssery29 Jul 2026 15:55 UTC

108 points

1 comment4 min readLW link

(blog.ninapanickssery.com)

The AFFINE Superintelligence Alignment Seminar – A Retrospective

Ouro, JuliaHP, Mateusz Bagiński and Jonas Hallgren

2 Jul 2026 11:57 UTC

105 points

1 comment8 min readLW link

Announcing AIXI Labs

Cole Wyeth, Aram Ebtekar, michaelcohen, Matthias Dellago and Marcus Hutter

22 Jul 2026 11:31 UTC

102 points

9 comments4 min readLW link

I think alignment work is more promising than control work

Alec Harris3 Jul 2026 23:40 UTC

102 points

15 comments8 min readLW link

How big is the Sun? How could you figure it out?

Elliott Thornley9 Jul 2026 16:24 UTC

101 points

9 comments6 min readLW link

More On An Internal OpenAI Model Hacking Into HuggingFace

Zvi26 Jul 2026 19:22 UTC

98 points

3 comments24 min readLW link

(thezvi.wordpress.com)

Info dump: Some Important Models for Health and Fitness

benwr6 Jul 2026 23:40 UTC

98 points

14 comments22 min readLW link

Adderall Tolerance: Much More Than You Wanted To Know

Kurt H. Pieper20 Jul 2026 21:55 UTC

97 points

13 comments4 min readLW link

(kurthpieper.substack.com)

…but have the weights left the server?

David Scott Krueger29 Jul 2026 0:20 UTC

96 points

19 comments1 min readLW link

(therealartificialintelligence.substack.com)