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 Jul AugSep

All 1 234 5 6 7 8 9 10 11 12 13 14 15 16

What is neuralese and why is it bad?

Linch2 Sep 2026 23:37 UTC

49 points

3 comments3 min readLW link

(linch.substack.com)

Notes on “A global workspace in language models”

Shunk2 Sep 2026 23:12 UTC

7 points

0 comments5 min readLW link

BLOOM-WILT: on-policy examples of any LLM behaviour, from a one-line description and logits alone

AdriansSkapars2 Sep 2026 23:11 UTC

7 points

0 comments10 min readLW link

Releasing Benign Trajectories for MonitoringBench

Can Küçükkurt and ethanelasky

2 Sep 2026 23:06 UTC

9 points

0 comments4 min readLW link

A key bottleneck for effective AI governance

Milly Medén2 Sep 2026 23:02 UTC

19 points

3 comments5 min readLW link

The Return of Aspect Oriented Programming

thomascolthurst2 Sep 2026 22:22 UTC

5 points

0 comments4 min readLW link

The Three P’s Theory of Research Organizations

thomascolthurst2 Sep 2026 22:19 UTC

5 points

0 comments2 min readLW link

Talking to journalists

KatjaGrace2 Sep 2026 21:52 UTC

36 points

21 comments3 min readLW link

(worldspiritsockpuppet.substack.com)

A proposal for a highly effective AI safety org

ceselder2 Sep 2026 21:34 UTC

72 points

14 comments2 min readLW link

Kairos has raised $50M to build talent infrastructure for AI safety (and we’re hiring!)

agucova2 Sep 2026 20:13 UTC

94 points

8 comments8 min readLW link

How concerned should we be about Astra’s recurrent architecture?

Rauno Arike2 Sep 2026 18:28 UTC

170 points

28 comments10 min readLW link

Two Paths from Logical Induction: Part 0

Ashe Vazquez Nuñez2 Sep 2026 17:53 UTC

32 points

0 comments12 min readLW link

Why OpenAI’s Astra Could Make AI Doom Harder to Prevent

Goutham Nalagatla2 Sep 2026 16:35 UTC

17 points

5 comments7 min readLW link

Life is not a social deduction game

Fernand02 Sep 2026 16:23 UTC

9 points

0 comments2 min readLW link

The opposite of culty isn’t bland

Fernand02 Sep 2026 15:28 UTC

3 points

0 comments3 min readLW link

Fiduciary Overlays for AI—with an RFP opportunity

aaguirre2 Sep 2026 15:13 UTC

11 points

0 comments1 min readLW link

Autonomy, Freedom and Control

Samuel Ratnam2 Sep 2026 13:35 UTC

5 points

0 comments4 min readLW link

(samuelratnam.substack.com)

Anthropic Has Some Alignment Problems

Zvi2 Sep 2026 13:21 UTC

40 points

3 comments14 min readLW link

(thezvi.wordpress.com)

If you’re interpreting <1B parameter models, you should use a tensor transformer

Logan Riggs2 Sep 2026 13:20 UTC

43 points

0 comments2 min readLW link

[Diagram] Early handoff? Improve conceptual reasoning?

Cleo Nardo2 Sep 2026 12:35 UTC

27 points

2 comments10 min readLW link

(clattubato.substack.com)

AI Co-operation and AI Alignment

drnickbone2 Sep 2026 11:19 UTC

21 points

1 comment2 min readLW link

Core Assumptions of ELK

Q Home2 Sep 2026 10:04 UTC

13 points

0 comments5 min readLW link

Applications Open for Impact Accelerator Program

High Impact Professionals2 Sep 2026 9:30 UTC

1 point

0 comments3 min readLW link

The economics of filling a lake with desalinated water

Yair Halberstadt2 Sep 2026 7:04 UTC

34 points

0 comments2 min readLW link

Incoherent AI Identities can also be Stable

Ashe Vazquez Nuñez2 Sep 2026 2:21 UTC

35 points

0 comments13 min readLW link

Don’t be the vitamin B guy

HedonicEscalator2 Sep 2026 2:02 UTC

68 points

6 comments6 min readLW link

(hedonicescalator.substack.com)

A Semitechnical Interlude on the Equivalence of Probabilistic and Deterministic Solomonoff Induction

N.C. Young2 Sep 2026 2:01 UTC

8 points

1 comment10 min readLW link

(ncyoung.substack.com)

Send me the prompt

dan.parshall2 Sep 2026 1:57 UTC

22 points

6 comments2 min readLW link

(canaryinstitute.substack.com)

For all available misaligned actions, expected punishment must exceed expected reward.

Jackson Hurley2 Sep 2026 1:15 UTC

1 point

0 comments1 min readLW link

Explaining Knightianism on one foot

Richard_Ngo2 Sep 2026 0:53 UTC

191 points

24 comments11 min readLW link

(www.mindthefuture.info)

Whistle Synth Mac App

jefftk2 Sep 2026 0:30 UTC

10 points

0 comments3 min readLW link

(www.jefftk.com)

Quantifying CoT faithfulness

Mihir Sahasrabudhe1 Sep 2026 23:15 UTC

3 points

0 comments5 min readLW link

“Collusion” is just cooperation that you don’t like

Morgan S1 Sep 2026 22:04 UTC

22 points

15 comments4 min readLW link

What happened to impact markets?

Austin Chen1 Sep 2026 21:30 UTC

23 points

3 comments3 min readLW link

A nightwatchman on every probe: Superintelligent surveillance to prevent galactic anarchy

Mia Taylor and Oscar

1 Sep 2026 21:04 UTC

16 points

0 comments7 min readLW link

(www.forethought.org)

No Bush, No Potato

Fernand01 Sep 2026 20:49 UTC

9 points

0 comments2 min readLW link

The Alignment Journal: Organization, Personnel, and Scope

Dan MacKinlay, JessRiedel, Daniel Murfet and Kristi Uustalu

1 Sep 2026 20:49 UTC

95 points

5 comments11 min readLW link

(blog.alignmentjournal.org)

Fake voices: warping the social world

KatjaGrace1 Sep 2026 20:49 UTC

33 points

1 comment3 min readLW link

(worldspiritsockpuppet.substack.com)

You can rarely pet the dog in an LLM-generated game

yamike1 Sep 2026 19:43 UTC

14 points

0 comments3 min readLW link

(mikeushakov.com)

Evolution could encode the brain in DNA

olehif1 Sep 2026 18:23 UTC

10 points

11 comments1 min readLW link

Research scientists for CaML: Investigating whether mid-training can survive RL

Jasmine Brazilek1 Sep 2026 18:13 UTC

16 points

2 comments2 min readLW link

Towards deployment-time misalignment continuation evals: lessons from recent loss of control incidents

Cath Ge-Wang1 Sep 2026 17:45 UTC

22 points

0 comments12 min readLW link

(cathgewang.substack.com)

Lumina Probiotic, Past and Future

jayterwahl1 Sep 2026 17:26 UTC

13 points

0 comments5 min readLW link

Lumina Probiotic Response to Critics

jayterwahl1 Sep 2026 17:25 UTC

2 points

3 comments5 min readLW link

New HPMoR Podcast site

Eneasz1 Sep 2026 17:13 UTC

19 points

0 comments1 min readLW link

A Year of Atheism

Laiba Rehman ✦ RJ1 Sep 2026 16:19 UTC

62 points

4 comments8 min readLW link

PauseAI Has ‘officially disendorsed’ PauseAI-US

nem1 Sep 2026 15:51 UTC

279 points

216 comments8 min readLW link

Best Way To Start a Local Group?

nem1 Sep 2026 15:44 UTC

12 points

2 comments1 min readLW link

Could internal model transparency tame the AI race?

Karthik Tadepalli1 Sep 2026 15:41 UTC

7 points

1 comment6 min readLW link

(blog.karthiktadepalli.com)

HuggingFace Attack Postmortem: Civilizations, Reactions and Next Actions

Zvi1 Sep 2026 14:10 UTC

45 points

10 comments55 min readLW link

(thezvi.wordpress.com)