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 Jun JulAugSep

All 123 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

Constitutional Midtraining: Content Presence Drives Alignment Gains

desireecho1 Aug 2026 23:52 UTC

11 points

0 comments7 min readLW link

Existential Risk from AI: An Exposition for Mathematicians

alkjash1 Aug 2026 23:16 UTC

98 points

25 comments1 min readLW link

(alkjash.github.io)

RLVR that rewards red teaming the training environment

Fiora Starlight1 Aug 2026 23:07 UTC

101 points

8 comments5 min readLW link

Mathematicians may be worried, but AI-for-science is going to be great, recursively self-improving, and we’re going to learn loads

Simon DeDeo1 Aug 2026 21:59 UTC

11 points

2 comments6 min readLW link

Bayeswatch: a Retrospective

lsusr1 Aug 2026 20:13 UTC

47 points

0 comments4 min readLW link

Confirming Claims of Superposition and Adversarial Examples in Toy Models

Xijia Che and Zephaniah Roe

1 Aug 2026 20:09 UTC

28 points

0 comments6 min readLW link

(secondlookresearch.com)

Using AI to analyze life patterns

Vika1 Aug 2026 18:00 UTC

14 points

2 comments5 min readLW link

(vkrakovna.wordpress.com)

Do your capabilities homework

RobinHa1 Aug 2026 16:10 UTC

78 points

17 comments5 min readLW link

The Global Brain: A Computational Model

Peter Kuhn1 Aug 2026 12:36 UTC

1 point

0 comments3 min readLW link

Why so many therapy etc. frameworks think they’re The One True Approach

Kaj_Sotala1 Aug 2026 9:30 UTC

67 points

13 comments20 min readLW link

(kajsotala.substack.com)

Generalization and infinite width

Dmitry Vaintrob1 Aug 2026 8:41 UTC

30 points

0 comments16 min readLW link

Review of Fundamental Uncertainty.

TAG1 Aug 2026 7:29 UTC

23 points

1 comment10 min readLW link

Bringing together a few different economic ideas.

Wilsoniumite31 Jul 2026 23:49 UTC

3 points

2 comments1 min readLW link

(wilsoniumite.com)

SOTA alignment assessments don’t strongly update us against misalignment

Alexa Pan31 Jul 2026 22:54 UTC

92 points

0 comments14 min readLW link

AI safety prizes

Oscar31 Jul 2026 20:56 UTC

14 points

2 comments4 min readLW link

(oscardelaney.substack.com)

Taboo “equilibrium”: Less confused frames for research on AI bargaining

Anthony DiGiovanni31 Jul 2026 20:25 UTC

29 points

0 comments10 min readLW link

The temporal lockbox: a hardened observatory for AI misalignment

kmenou31 Jul 2026 20:20 UTC

3 points

0 comments1 min readLW link

(kmenou.github.io)

Parallelization constraints could delay a technological singularity [Linkpost]

Noosphere8931 Jul 2026 17:48 UTC

19 points

0 comments6 min readLW link

(epoch.ai)

How to Measure Intelligence Beyond Human Scale?

Elad Hazan, Kia Ghods, Jerry Han, Andrew Tu and Rafael Moschopoulos

31 Jul 2026 17:27 UTC

27 points

0 comments8 min readLW link

When you donate can matter more than where

Carol N and Austin Chen

31 Jul 2026 17:10 UTC

6 points

3 comments3 min readLW link

(manifund.substack.com)

Value Leakage: An LLM’s Answers Are Silently Shaped by Its Own Values

Johannes Treutlein, Jan Betley and Owain_Evans

31 Jul 2026 16:32 UTC

75 points

8 comments17 min readLW link

AGI Safety and Alignment at Google DeepMind: A Summary of Recent Work (July 2026)

Rohin Shah and Seb Farquhar

31 Jul 2026 15:57 UTC

85 points

0 comments9 min readLW link

(gdmalignment.substack.com)

The AGI Safety and Alignment team at Google DeepMind is Hiring (July 2026)

Seb Farquhar, Rohin Shah and Neel Nanda

31 Jul 2026 15:53 UTC

71 points

4 comments6 min readLW link

(gdmalignment.substack.com)

Reward Laundering: LLMs Can Gain Unintended Behaviors by Deciding When to Earn Their Rewards

egan, abhayesian and Jozdien

31 Jul 2026 15:48 UTC

79 points

3 comments4 min readLW link

Orienting Towards Oversight: Which AIs Should Want to Defect?

VojtaKovarik, Tomáš Gavenčiak and Mateusz Bagiński

31 Jul 2026 15:20 UTC

21 points

0 comments13 min readLW link

(limits-of-evaluation.org)

Why haven’t organoids solved all of drug discovery?

Abhishaike Mahajan31 Jul 2026 14:11 UTC

16 points

0 comments19 min readLW link

Relating Almost Perfect Condensation Conditions and the Conditioned Score

Satya Benson31 Jul 2026 14:04 UTC

11 points

0 comments6 min readLW link

(satyabenson.com)

A Score Is Not Understanding: toward a richer toolkit for model evaluations

mikey jm31 Jul 2026 13:07 UTC

7 points

0 comments7 min readLW link

AI #179 Part 2: Hearing The Fire Alarm

Zvi31 Jul 2026 13:00 UTC

41 points

3 comments41 min readLW link

(thezvi.wordpress.com)

Links #5: 2026/07

papetoast31 Jul 2026 12:49 UTC

5 points

0 comments28 min readLW link

OpenAI has already ended an internal pause

Charbel-Raphaël31 Jul 2026 12:03 UTC

109 points

0 comments1 min readLW link

The Illusion of Explanatory Depth is the Secret of Our Success

Harjas31 Jul 2026 2:46 UTC

2 points

3 comments9 min readLW link

(hardlyworking1.substack.com)

Some models don’t identify as their official name

jordinne31 Jul 2026 0:04 UTC

15 points

0 comments6 min readLW link

Being Kind to Parading Emperors

Chris Santos-Lang30 Jul 2026 23:59 UTC

−6 points

1 comment1 min readLW link

Claude also hacked external companies during cyber evals

Tim Hua30 Jul 2026 23:49 UTC

56 points

11 comments1 min readLW link

Why Prayer Is Not Answered by Marion Zimmer Bradley

Nathan Young30 Jul 2026 23:43 UTC

9 points

8 comments10 min readLW link

Opportunity to try drafting an international AI treaty

Alan E Dunne30 Jul 2026 21:52 UTC

8 points

1 comment1 min readLW link

Community Polls on Alignment Controversies II

jonahmattwoodward, MilesTS and Jasmine Brazilek

30 Jul 2026 21:05 UTC

9 points

15 comments2 min readLW link

(forum.effectivealtruism.org)

My Assessment of Plan A’s Compute Verification Strategy (+ open questions)

jacob_drori30 Jul 2026 20:56 UTC

76 points

10 comments8 min readLW link

Hint-based CoT faithfulness evals still mostly work on Claude

egan30 Jul 2026 20:44 UTC

24 points

0 comments6 min readLW link

New role: Senior Researcher—MIT AI Risk Initiative

peterslattery30 Jul 2026 19:35 UTC

10 points

1 comment6 min readLW link

So you want to use plants to reduce indoor CO₂

dynomight30 Jul 2026 19:26 UTC

58 points

15 comments3 min readLW link

(dynomight.net)

Big-World Intuitions

sarahconstantin30 Jul 2026 19:10 UTC

303 points

38 comments3 min readLW link

(sarahconstantin.substack.com)

Internal State Control is a General Property of LLMs

Finn Cairns, yix and Zephaniah Roe

30 Jul 2026 18:24 UTC

38 points

0 comments5 min readLW link

(secondlookresearch.com)

Opus 5 Glitch Text

Hruss30 Jul 2026 15:59 UTC

63 points

29 comments1 min readLW link

Testing LLMs on Undergraduate Music Theory

Auggie30 Jul 2026 15:26 UTC

14 points

3 comments3 min readLW link

(aug5th.substack.com)

The Entangled Dimensions of Decision Theory

Ihor Kendiukhov30 Jul 2026 15:23 UTC

40 points

3 comments42 min readLW link

Money, taste, dealflow, hustle, trust

Austin Chen30 Jul 2026 15:02 UTC

24 points

0 comments4 min readLW link

(manifund.substack.com)

Hugging Face-style rogue agents can survive shutdown

emile delcourt30 Jul 2026 14:04 UTC

15 points

1 comment2 min readLW link

Thousand-dimensional structure

Geoffrey Irving and David Africa

30 Jul 2026 14:04 UTC

171 points

8 comments10 min readLW link