4 Jan 2026 22:11 UTC

146 points

10 comments142 min readLW link

LessOnline 2026 Improvement Ideas

nomagicpill4 Jan 2026 21:56 UTC

16 points

0 comments1 min readLW link

The economy is a graph, not a pipeline

anithite4 Jan 2026 21:48 UTC

33 points

10 comments4 min readLW link

Calling all college students (and new readers)

neo4 Jan 2026 21:20 UTC

15 points

0 comments1 min readLW link

Rock bottom terminal value

ihatenumbersinusernames74 Jan 2026 20:43 UTC

4 points

9 comments2 min readLW link

In My Misanthropy Era

jenn4 Jan 2026 18:34 UTC

352 points

153 comments8 min readLW link

(jenn.site)

The Thinking Machine

PeterMcCluskey4 Jan 2026 18:24 UTC

36 points

0 comments2 min readLW link

(bayesianinvestor.com)

The Maduro Polymarket bet is not “obviously insider trading”

ceselder4 Jan 2026 10:53 UTC

22 points

18 comments3 min readLW link

The Problem with Democracy

RandStrauss4 Jan 2026 7:11 UTC

−3 points

3 comments2 min readLW link

Examples of Subtle Alignment Failures from Claude and Gemini

Tachikoma4 Jan 2026 4:29 UTC

−9 points

1 comment5 min readLW link

Four Downsides of Training Policies Online

Alek Westover and egan

4 Jan 2026 3:17 UTC

29 points

4 comments3 min readLW link

Humanity’s Gambit

Ben Ihrig4 Jan 2026 3:08 UTC

5 points

5 comments3 min readLW link

Semantic Topological Spaces

TristanTrim4 Jan 2026 0:58 UTC

11 points

16 comments5 min readLW link

The surprising adequacy of the Roblox game marketplace

Esteban Restrepo3 Jan 2026 14:15 UTC

26 points

3 comments8 min readLW link

(papabos.substack.com)

Re: Anthropic Chinese Cyber-Attack. How Do We Protect Open-source Models?

Mayowa Osibodu3 Jan 2026 9:45 UTC

−1 points

2 comments6 min readLW link

Give Skepticism a Try

Ape in the coat3 Jan 2026 8:57 UTC

12 points

17 comments3 min readLW link

(apeinthecoat102771.substack.com)

Why We Should Talk Specifically Amid Uncertainty

sbaumohl3 Jan 2026 3:04 UTC

11 points

1 comment7 min readLW link

Companies as “proto-ASI”

beyarkay (Boyd Kane)3 Jan 2026 0:24 UTC

15 points

3 comments1 min readLW link

(boydkane.com)

AXRP Episode 47 - David Rein on METR Time Horizons

DanielFilan3 Jan 2026 0:10 UTC

21 points

0 comments46 min readLW link

The Weirdness of Dating/Mating: Deep Nonconsent Preference

johnswentworth2 Jan 2026 23:05 UTC

12 points

61 comments6 min readLW link

Can AI learn human societal norms from social feedback (without recapitulating all the ways this has failed in human history?)

foodforthought2 Jan 2026 22:11 UTC

7 points

3 comments4 min readLW link

Fertility Roundup #5: Causation

Zvi2 Jan 2026 22:00 UTC

19 points

5 comments25 min readLW link

(thezvi.wordpress.com)

Scale-Free Goodness

testingthewaters2 Jan 2026 21:00 UTC

10 points

3 comments5 min readLW link

(aclevername.substack.com)

Does developmental cognitive psychology provide any hints for making model alignment more robust?

foodforthought2 Jan 2026 20:31 UTC

7 points

0 comments3 min readLW link

Does evolution provide any hints for making model alignment more robust?

foodforthought2 Jan 2026 19:06 UTC

5 points

0 comments4 min readLW link

Where do AI Safety Fellows go? Analyzing a dataset of 600+ alumni

Christopher_Clay2 Jan 2026 18:14 UTC

20 points

2 comments5 min readLW link

(forum.effectivealtruism.org)

Instruct Vectors—Base models can be instruct with activation vectors

Eriskii2 Jan 2026 18:14 UTC

21 points

0 comments8 min readLW link

[Advanced Intro to AI Alignment] 2. What Values May an AI Learn? — 4 Key Problems

Towards_Keeperhood2 Jan 2026 14:51 UTC

33 points

10 comments19 min readLW link

2025 Letter

zef2 Jan 2026 13:57 UTC

10 points

0 comments14 min readLW link

(zephyyr.substack.com)

2025 in AI predictions

jessicata2 Jan 2026 4:29 UTC

245 points

19 comments11 min readLW link

Debunking claims about subquadratic attention

Vladimir Ivanov2 Jan 2026 4:23 UTC

32 points

5 comments3 min readLW link

The bio-pirate’s guide to GLP-1 agonists

quiet_NaN2 Jan 2026 3:32 UTC

40 points

3 comments5 min readLW link

College Was Not That Terrible Now That I’m Not That Crazy

Zack_M_Davis1 Jan 2026 23:14 UTC

90 points

9 comments44 min readLW link

(zackmdavis.net)

Taiwan war timelines might be shorter than AI timelines

Baram Sosis1 Jan 2026 22:30 UTC

108 points

21 comments5 min readLW link

Split (Part 1)

Shoshannah Tekofsky1 Jan 2026 22:29 UTC

27 points

2 comments4 min readLW link

(shoshanigans.substack.com)

[Question] Who is responsible for shutting down rogue AI?

Cole Wyeth1 Jan 2026 21:36 UTC

45 points

2 comments1 min readLW link

$500 Write like lsusr competition—Results

lsusr1 Jan 2026 20:53 UTC

40 points

4 comments3 min readLW link

Overwhelming Superintelligence

Raemon1 Jan 2026 20:51 UTC

80 points

30 comments1 min readLW link

Reducing MDMA neurotoxicity

Pjain1 Jan 2026 20:13 UTC

5 points

0 comments12 min readLW link

Is it possible to prevent AGI?

jrincayc1 Jan 2026 19:15 UTC

12 points

1 comment2 min readLW link

Principled Interpretability of Reward Hacking in Closed Frontier Models

gersonkroiz, aditya singh, Senthooran Rajamanoharan and Neel Nanda

1 Jan 2026 16:37 UTC

24 points

0 comments23 min readLW link

AI #149: 3

Zvi1 Jan 2026 15:40 UTC

39 points

7 comments23 min readLW link

(thezvi.wordpress.com)

ML Engineer—MIT AI Risk Initiative, Contractor, Part-time, 6-months

peterslattery1 Jan 2026 14:23 UTC

4 points

0 comments1 min readLW link

Recent LLMs can do 2-hop and 3-hop latent (no-CoT) reasoning on natural facts

ryan_greenblatt1 Jan 2026 13:36 UTC

129 points

11 comments3 min readLW link

AGI and the structural foundations of democracy and the rule-based international order

PabloAMC1 Jan 2026 12:07 UTC

21 points

0 comments10 min readLW link

(pabloamc.substack.com)

From Drift to Snap: Instruction Violation as a Phase Transition

James Hoffend1 Jan 2026 10:44 UTC

8 points

0 comments3 min readLW link

Quick polls on AGI doom

denkenberger1 Jan 2026 6:23 UTC

2 points

0 comments1 min readLW link

(forum.effectivealtruism.org)

Special Persona Training: Hyperstition Progress Report 2

jayterwahl1 Jan 2026 1:34 UTC

38 points

2 comments2 min readLW link

You will be OK

Boaz Barak1 Jan 2026 0:33 UTC

57 points

57 comments4 min readLW link

Speciesquest 2026

eukaryote31 Dec 2025 23:24 UTC

27 points

3 comments5 min readLW link

(eukaryotewritesblog.com)