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 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 171819 20 21 22 23 24 25 26 27 28 29 30 31

You Can’t Iterate to Trustworthy AI Code Without Understanding

ronbodkin17 Aug 2026 23:00 UTC

8 points

0 comments10 min readLW link

Do your readers get annoyed if you post every day for 30 days?

Steff17 Aug 2026 21:59 UTC

13 points

5 comments1 min readLW link

What 27 AI Safety Generalists Are Building This Summer

agucova, Jesse Gilbert and Helena Tran

17 Aug 2026 21:56 UTC

18 points

0 comments9 min readLW link

What gives you away: how LLMs form opinions of you

Cat McGee17 Aug 2026 20:59 UTC

42 points

8 comments6 min readLW link

Evaluating Chain-of-Thought Monitorability is Still an Open Problem: Comments on OpenAI’s Monitorability Evals

Connor Dilgren and Sarah Wiegreffe

17 Aug 2026 19:45 UTC

20 points

3 comments18 min readLW link

Weird Re-Tokenization, Symmetries and Compression: Research Agenda

Xenomirant and Sami Wolf

17 Aug 2026 19:43 UTC

21 points

15 comments16 min readLW link

Fork Around and Find Out Part 3: Interpreting the knight auditor

David Litman17 Aug 2026 18:11 UTC

18 points

0 comments5 min readLW link

Connect to your future selves

PatrickDFarley17 Aug 2026 14:57 UTC

14 points

0 comments11 min readLW link

Lateral Workshop Applications: A Case Study

Jacob Brinton and Seth Lifland

17 Aug 2026 10:38 UTC

10 points

0 comments3 min readLW link

(forum.effectivealtruism.org)

Value Alignment Is a Pseudo Concept. A translation; humanity is not a single subject, and alignment is not one-way.

Davidmanheim17 Aug 2026 7:44 UTC

36 points

1 comment10 min readLW link

Untie Squared ReLU variant

Amy_17 Aug 2026 6:15 UTC

15 points

5 comments7 min readLW link

Study Update: Does post-training quantization change welfare-relevant indicators in open-weight language models?

ashesfall16 Aug 2026 21:11 UTC

8 points

0 comments7 min readLW link

Q2.5 2026 Timelines Update: Uplift and Revenue

brendanhalstead, Daniel Kokotajlo and elifland

16 Aug 2026 19:00 UTC

66 points

20 comments11 min readLW link

(blog.aifutures.org)

Case for Funding AI Safety in Japan

mmKALLL16 Aug 2026 18:41 UTC

33 points

1 comment9 min readLW link

Will There Be an AI Hegemon? A Mental Model for AI Power Concentration

simeon_c16 Aug 2026 18:41 UTC

28 points

7 comments3 min readLW link

(simeoncampos.substack.com)

The Doomsday Argument is Reasonable and Mostly Points to Longevity

Josh Snider16 Aug 2026 17:14 UTC

29 points

12 comments6 min readLW link

Three thoughts on civilisational handoff

Cleo Nardo16 Aug 2026 17:12 UTC

62 points

2 comments2 min readLW link

(clattubato.substack.com)

Should Less Wrong add subtitles?

Chris_Leong16 Aug 2026 9:05 UTC

32 points

10 comments1 min readLW link

Are questions allowed on LessWrong?

yatharth16 Aug 2026 7:04 UTC

13 points

10 comments2 min readLW link

Does DiffusionGemma do latent reasoning?

Jan Bauer and Neel Nanda

16 Aug 2026 4:22 UTC

40 points

0 comments9 min readLW link

Winners of the Fundamental Uncertainty Essay Contest

Gordon Seidoh Worley16 Aug 2026 1:20 UTC

27 points

2 comments1 min readLW link

(www.uncertainupdates.com)

Kimi likes causal decision theory more after RL in twin prisoner’s dilemmas

oakhu15 Aug 2026 22:31 UTC

115 points

70 comments6 min readLW link

Luck is a function of surface area.

sid.the.manne@gmail.com15 Aug 2026 20:44 UTC

11 points

0 comments1 min readLW link

What if Parameter Updates were Text?

DaemonicSigil15 Aug 2026 20:06 UTC

27 points

0 comments11 min readLW link

Using Chunked Monitoring to Detect Deception in Long Transcripts

sfereido15 Aug 2026 20:04 UTC

11 points

0 comments4 min readLW link

How To Catch a Distilled Model

Goutham Nalagatla and Carlos Guerrero Alvarez

15 Aug 2026 19:09 UTC

20 points

8 comments9 min readLW link

Mom’s Advice For Hosting A Class Reunion

jenn15 Aug 2026 15:52 UTC

53 points

2 comments2 min readLW link

Learning new facts can change LLM behaviour

Richard Juggins15 Aug 2026 13:46 UTC

33 points

2 comments14 min readLW link

(www.workingthroughai.com)

On Dwarkesh Patel’s Podcast With Ryan Greenblatt

Zvi15 Aug 2026 13:00 UTC

34 points

1 comment29 min readLW link

(thezvi.wordpress.com)

Does doubling a user profile change the effect of an instruction about using saved memories?

theodorepjs15 Aug 2026 11:59 UTC

17 points

1 comment3 min readLW link

I’m starting a interview series of people working in Lean / formal methods / math formalization

Adi Baradwaj15 Aug 2026 9:15 UTC

10 points

0 comments1 min readLW link

Nuclear physics of Alex Zhao’s comment for “Pacing the Frontier”

Dante Dam15 Aug 2026 9:03 UTC

17 points

0 comments11 min readLW link

Rerunning AI safety papers on every frontier release would be pretty easy and valuable

Zephaniah Roe, hersheys and yix

15 Aug 2026 5:14 UTC

123 points

9 comments4 min readLW link

(secondlookresearch.com)

Metaphilosophy II: Empirical Flywheels

interstice15 Aug 2026 2:08 UTC

18 points

12 comments5 min readLW link

(thermontology.com)

Red vs Blue, but for Evals

Morgan S15 Aug 2026 1:11 UTC

20 points

1 comment16 min readLW link

Toy Model of Activation Obfuscation

Jesse Li14 Aug 2026 23:25 UTC

16 points

0 comments7 min readLW link

(jesseli2002.github.io)

Your Agents Are Not Time Aware

Michael Ofengenden and Maksym Andriushchenko

14 Aug 2026 23:17 UTC

40 points

4 comments11 min readLW link

Announcing: Iliad’s New 2026 Fellowships

David Udell, Alexander Gietelink Oldenziel and Leon Lang

14 Aug 2026 22:41 UTC

37 points

0 comments1 min readLW link

Training a Conceptual Reasoning Judge

Kevin Zhang and Caspar Oesterheld

14 Aug 2026 20:33 UTC

17 points

1 comment6 min readLW link

Why I’m Skeptical of Longtermism

James Brobin14 Aug 2026 20:17 UTC

1 point

13 comments3 min readLW link

Empty Chambers, Missing Stairs

finitude14 Aug 2026 19:37 UTC

8 points

0 comments5 min readLW link

Do It Like Darwin

derelict543214 Aug 2026 19:11 UTC

14 points

5 comments7 min readLW link

The Day Humanity Died (Parody of American Pie)

Bentham's Bulldog14 Aug 2026 18:27 UTC

8 points

1 comment3 min readLW link

Scrying, Modeling, and Nerdsnipe

Cole Wyeth14 Aug 2026 17:59 UTC

76 points

2 comments5 min readLW link

Who we’d like as regrantors

Carol N and Austin Chen

14 Aug 2026 17:40 UTC

8 points

0 comments4 min readLW link

(manifund.substack.com)

How the American Executive Could Control AI Companies

caiitlinm and Anders Cairns Woodruff

14 Aug 2026 15:43 UTC

73 points

0 comments13 min readLW link

Open Problems in Mechanistic Intepretability of Biological AIs

Ihor Kendiukhov14 Aug 2026 15:35 UTC

21 points

0 comments2 min readLW link

V&V takes on “Pacing the frontier”

Yoav Hollander14 Aug 2026 13:25 UTC

4 points

2 comments9 min readLW link

(blog.foretellix.com)

Avoiding the corporate treacherous turn: crowd-sourcing design ideas

Stuart_Armstrong14 Aug 2026 13:10 UTC

26 points

14 comments1 min readLW link

Frontier agents don’t comply with standards, even when instructed to

Daan Henselmans and Arno Libert

14 Aug 2026 12:08 UTC

42 points

5 comments7 min readLW link