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

All 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 222324 25 26 27 28 29 30 31

A Multi-Agent Extension for Petri

carissacullen22 Jul 2026 21:51 UTC

11 points

0 comments4 min readLW link

In other words: The influence of prompt variation on alignment evals

lwroe, AJ Weeks and morio

22 Jul 2026 21:51 UTC

22 points

0 comments7 min readLW link

Two Coefficient Givings beat one twice as big

jackultraphil22 Jul 2026 21:46 UTC

17 points

0 comments12 min readLW link

Can an LLM make a feature-length movie on its own?

Josh Snider22 Jul 2026 20:13 UTC

49 points

11 comments4 min readLW link

Will almost all future companies eventually be founded and run by autonomous AIs?

Steven Byrnes22 Jul 2026 20:04 UTC

86 points

4 comments8 min readLW link

OpenAI Model Hacks Into HuggingFace During Cybersecurity Evaluation

Zvi22 Jul 2026 19:31 UTC

91 points

6 comments25 min readLW link

(thezvi.wordpress.com)

We cannot simulate AI security research

Jafar Isbarov22 Jul 2026 19:15 UTC

8 points

0 comments5 min readLW link

The Conjecture of Strong Subjectivity

D.Schetselaar22 Jul 2026 17:43 UTC

9 points

18 comments1 min readLW link

(philpapers.org)

The Best AI Bill Congress Hasn’t Introduced Yet

dan.parshall22 Jul 2026 16:19 UTC

11 points

0 comments11 min readLW link

Your AIs don’t do what you want. This is really bad

Kaustubh Kislay22 Jul 2026 15:56 UTC

16 points

0 comments3 min readLW link

(rewardhacking.org)

Models don’t seem to be dishonest in the way humans are

David Africa and Jacob Pfau

22 Jul 2026 15:32 UTC

46 points

4 comments9 min readLW link

Mechanistic interpretability hypotheses for Measuring Reward-Seeking by Instilling Contrastive Beliefs and additional comments

Burny22 Jul 2026 14:58 UTC

16 points

0 comments6 min readLW link

How do neural networks regress cubic polynomials? Apparently, they use a trick invented in Milan 500 years ago

enricobottazzi22 Jul 2026 14:42 UTC

29 points

0 comments16 min readLW link

(2/3) The Dangers of AGI

Eigenbraid22 Jul 2026 13:57 UTC

9 points

0 comments9 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

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

[Paper] Stringological sequence prediction II

Vanessa Kosoy22 Jul 2026 7:27 UTC

41 points

0 comments1 min readLW link

(arxiv.org)

OpenAI and Hugging Face partner to address security incident during model evaluation

Matrice Jacobine22 Jul 2026 6:30 UTC

25 points

1 comment1 min readLW link

(openai.com)

OpenAI Models Behind HuggingFace Cybersecurity Incident

LawrenceC21 Jul 2026 21:36 UTC

220 points

20 comments1 min readLW link

WeirdChat: A catalog of unexpected AI behaviors, discovered automatically

neilchowdhury21 Jul 2026 21:12 UTC

63 points

4 comments10 min readLW link

(transluce.org)

OpenAI Shares Some Alignment Problems

Zvi21 Jul 2026 19:41 UTC

149 points

8 comments9 min readLW link

(thezvi.wordpress.com)

Blogging Technology Interlude

Zack_M_Davis21 Jul 2026 19:03 UTC

14 points

1 comment3 min readLW link

(zackmdavis.net)

Ending Soon: Fundamental Uncertainty $2,000 Essay Contest

Gordon Seidoh Worley21 Jul 2026 16:51 UTC

15 points

0 comments1 min readLW link

(www.uncertainupdates.com)

Steering Blackmail Through a Model’s “Emotional State”

Chijioke Ugwuanyi and zahraa alsahili

21 Jul 2026 16:03 UTC

12 points

0 comments8 min readLW link

Measuring Reward-Seeking via Contrastive Belief Updates

Jérémy Scheurer, Axel Højmark, jenny, Felix Hofstätter, Theodore Ehrenborg, Bronson Schoen and Alex Meinke

21 Jul 2026 15:22 UTC

91 points

5 comments12 min readLW link

(rewardseeking.ai)

Measuring Reward-Seeking by Instilling Contrastive Beliefs

papetoast21 Jul 2026 15:11 UTC

12 points

0 comments1 min readLW link

(alignment.openai.com)

11 Open Empirical Problems in Reward-Seeking

Alex Meinke, Jérémy Scheurer, Axel Højmark and Theodore Ehrenborg

21 Jul 2026 15:08 UTC

65 points

0 comments8 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

Epistemics and Coordination: It’s complicated!

Raymond Douglas21 Jul 2026 12:23 UTC

54 points

13 comments8 min readLW link

(1/3) The Dangers of LLMs

Eigenbraid21 Jul 2026 12:01 UTC

10 points

0 comments6 min readLW link

I ran the standard AI litmus tests on my two toddlers (yep)

Carlo Valenti21 Jul 2026 11:38 UTC

45 points

8 comments4 min readLW link

Unofficial ACX/LW/EA Munich Bi-weekly Community Dinner

hilll21 Jul 2026 9:55 UTC

1 point

0 comments1 min readLW link

NameRank: The Model Knows Your Project, Not You.

Jarrett Ye21 Jul 2026 2:21 UTC

20 points

7 comments1 min readLW link

(01.me)

The Case for Physical AI Safety

kaylene_stocking and Bear Häon

20 Jul 2026 22:18 UTC

13 points

2 comments22 min readLW link

(paisi.ai)

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)

What do I mean by “Artificial General Intelligence”?

Steven Byrnes20 Jul 2026 21:13 UTC

44 points

5 comments4 min readLW link

AI 2040: Is it Actually a Deal?

1a3orn20 Jul 2026 20:58 UTC

55 points

13 comments9 min readLW link

Banana in, Bostrom out: paperclip maximization is one token-direction swap away (in Qwen 3.6-27B)

Jeffrey William Shorthill20 Jul 2026 20:36 UTC

8 points

4 comments4 min readLW link

The AI Safety Illusion: Why Current Safety Datasets Fool Us on Model Safety

Shahriar Golchin20 Jul 2026 20:12 UTC

5 points

0 comments8 min readLW link

Does routine compression undo LLM unlearning? A short project

hannahTao20 Jul 2026 20:08 UTC

20 points

0 comments3 min readLW link

Attempt at Finding Alignment Faking on Llama 70B to test sleeper-agent detection generalizes

skn873320 Jul 2026 20:04 UTC

9 points

1 comment6 min readLW link

Frontier AI lab misalignment risk, lessons from trading post-2008

Peter Chatwell20 Jul 2026 20:01 UTC

4 points

3 comments4 min readLW link

AI Voice Phishing Performs on Par With Human Scammers at a Fraction of the Cost

Fred Heiding and Simon Lermen

20 Jul 2026 19:59 UTC

28 points

3 comments6 min readLW link

The Infinity Fallacy

pvutov20 Jul 2026 19:57 UTC

1 point

5 comments1 min readLW link

Bluedot SF hackathon submission: A workable blueprint for a chip licensing regime

edpaulino20 Jul 2026 19:57 UTC

9 points

0 comments3 min readLW link

A requiem to the plan-vanilla Google search: On Sam Roweis, Mbappe and an Irish bar world cup watch party

Triadic dissclosures: A blog about unlikely triadic closures20 Jul 2026 19:56 UTC

5 points

0 comments4 min readLW link

Fable is SOTA at CIFAR Speedrun (& specification gaming)

rohuang, kaivu, zef and leni

20 Jul 2026 19:54 UTC

29 points

0 comments6 min readLW link

(fulcrum.inc)

Bluedot Technical AI Safety Puzzle, submission that got me a second place

perduta20 Jul 2026 19:53 UTC

13 points

0 comments8 min readLW link

(blog.perduta.net)

PauseCon London ’26: Applications now open

jonathan@pauseai20 Jul 2026 19:51 UTC

12 points

0 comments1 min readLW link

Restoring Model Alignment via Honesty Activation Steering

niklas_herbster, Martin Zborowski, Alberto Tosato, Gauthier Gidel and Tommaso (Tommie) Tosato

20 Jul 2026 19:51 UTC

18 points

0 comments12 min readLW link

(arxiv.org)