Eliezer and I wrote a book: If Any­one Builds It, Every­one Dies

So8resMay 14, 2025, 7:00 PM
594 points
99 comments2 min readLW link

Ori­ent­ing Toward Wizard Power

johnswentworthMay 8, 2025, 5:23 AM
504 points
128 comments5 min readLW link

In­ter­pretabil­ity Will Not Reli­ably Find De­cep­tive AI

Neel NandaMay 4, 2025, 4:32 PM
279 points
45 comments7 min readLW link

Too Soon

Gordon Seidoh WorleyMay 13, 2025, 3:01 PM
208 points
19 comments4 min readLW link

PSA: The LessWrong Feed­back Service

JustisMillsMay 12, 2025, 4:34 PM
203 points
12 comments2 min readLW link

What We Learned from Briefing 70+ Law­mak­ers on the Threat from AI

leticiagarciaMay 27, 2025, 6:23 PM
189 points
2 comments16 min readLW link
(substack.com)

Gem­ini Diffu­sion: watch this space

Yair HalberstadtMay 20, 2025, 7:29 PM
183 points
28 comments1 min readLW link
(deepmind.google)

Slow­down After 2028: Com­pute, RLVR Uncer­tainty, MoE Data Wall

Vladimir_NesovMay 1, 2025, 1:54 PM
172 points
22 comments5 min readLW link

Win­ning the power to lose

KatjaGraceMay 20, 2025, 6:40 AM
150 points
44 comments2 min readLW link
(worldspiritsockpuppet.com)

Con­sider not donat­ing un­der $100 to poli­ti­cal candidates

DanielFilanMay 11, 2025, 3:20 AM
132 points
32 comments1 min readLW link
(danielfilan.com)

AI Doomerism in 1879

David GrossMay 13, 2025, 2:48 AM
132 points
36 comments8 min readLW link

It’s Okay to Feel Bad for a Bit

moridinamaelMay 10, 2025, 11:24 PM
131 points
26 comments3 min readLW link

So­cial Anx­iety Isn’t About Be­ing Liked

ChipmonkMay 16, 2025, 10:26 PM
124 points
21 comments2 min readLW link
(chrislakin.blog)

Five Hinge‑Ques­tions That De­cide Whether AGI Is Five Years Away or Twenty

charlieoneillMay 6, 2025, 2:48 AM
124 points
17 comments5 min readLW link

Med­i­ta­tions on Doge

Martin SustrikMay 25, 2025, 12:00 PM
123 points
39 comments9 min readLW link
(250bpm.substack.com)

It’s hard to make schem­ing evals look re­al­is­tic for LLMs

May 24, 2025, 7:17 PM
117 points
16 comments5 min readLW link

One Year in DC

tlevinMay 19, 2025, 7:46 PM
110 points
5 commentsLW link
(www.greentape.pub)

UK AISI’s Align­ment Team: Re­search Agenda

May 7, 2025, 4:33 PM
109 points
2 comments11 min readLW link

What OpenAI Told Cal­ifor­nia’s At­tor­ney General

garrisonMay 17, 2025, 11:14 PM
108 points
3 commentsLW link
(www.obsolete.pub)

Notes on the Long Tasks METR pa­per, from a HCAST task contributor

abstractapplicMay 4, 2025, 11:17 PM
108 points
7 comments2 min readLW link

Please Donate to CAIP (Post 1 of 6 on AI Gover­nance)

Mass_DriverMay 7, 2025, 5:13 PM
106 points
20 comments33 min readLW link

Pri­ori­tiz­ing Work

jefftkMay 1, 2025, 2:00 AM
106 points
11 comments1 min readLW link
(www.jefftk.com)

AI Gover­nance to Avoid Ex­tinc­tion: The Strate­gic Land­scape and Ac­tion­able Re­search Questions

May 1, 2025, 10:46 PM
105 points
7 comments8 min readLW link
(techgov.intelligence.org)

If you’re not sure how to sort a list or grid—se­ri­ate it!

gwernMay 28, 2025, 3:54 AM
105 points
0 comments2 min readLW link
(www.jstatsoft.org)

We’re Not Ad­ver­tis­ing Enough (Post 3 of 6 on AI Gover­nance)

Mass_DriverMay 22, 2025, 5:05 PM
103 points
10 comments28 min readLW link

RA x Con­trolAI video: What if AI just keeps get­ting smarter?

WriterMay 2, 2025, 2:19 PM
100 points
17 comments9 min readLW link

The Ukraine War and the Kill Market

Martin SustrikMay 4, 2025, 7:50 AM
98 points
13 comments5 min readLW link
(250bpm.substack.com)

Gen­er­at­ing the Fun­niest Joke with RL (ac­cord­ing to GPT-4.1)

aggMay 16, 2025, 5:09 AM
96 points
22 comments4 min readLW link

a con­fu­sion about prefer­ence orderings

nostalgebraistMay 11, 2025, 7:30 PM
92 points
37 comments11 min readLW link

Slow cor­po­ra­tions as an in­tu­ition pump for AI R&D automation

May 9, 2025, 2:49 PM
91 points
23 comments9 min readLW link

The Sweet Les­son: AI Safety Should Scale With Compute

Jesse HooglandMay 5, 2025, 7:03 PM
89 points
1 comment3 min readLW link

Sea­son Re­cap of the Village: Agents raise $2,000

Shoshannah TekofskyMay 27, 2025, 12:34 PM
86 points
6 comments6 min readLW link
(theaidigest.org)

The Best Refer­ence Works for Every Subject

Parker ConleyMay 14, 2025, 12:58 AM
86 points
9 comments6 min readLW link
(parconley.com)

Dreams of Ideas

Joseph MillerMay 19, 2025, 2:15 AM
86 points
3 comments4 min readLW link

Claude 4 You: Safety and Alignment

ZviMay 25, 2025, 2:00 PM
82 points
7 comments63 min readLW link
(thezvi.wordpress.com)

As­so­ci­a­tion taxes are col­lu­sion subsidies

KatjaGraceMay 27, 2025, 6:50 AM
81 points
7 comments1 min readLW link
(worldspiritsockpuppet.com)

AIs at the cur­rent ca­pa­bil­ity level may be im­por­tant for fu­ture safety work

ryan_greenblattMay 12, 2025, 2:06 PM
80 points
2 comments4 min readLW link

The stakes of AI moral status

Joe CarlsmithMay 21, 2025, 6:20 PM
76 points
62 comments14 min readLW link
(joecarlsmith.substack.com)

Misal­ign­ment and Strate­gic Un­der­perfor­mance: An Anal­y­sis of Sand­bag­ging and Ex­plo­ra­tion Hacking

May 8, 2025, 7:06 PM
75 points
1 comment15 min readLW link

Policy recom­men­da­tions re­gard­ing re­pro­duc­tive technology

TsviBTMay 22, 2025, 2:49 PM
75 points
2 comments3 min readLW link

Sleep need re­duc­tion therapies

harsimonyMay 21, 2025, 3:22 PM
75 points
18 comments10 min readLW link
(splittinginfinity.substack.com)

FLF Fel­low­ship on AI for Hu­man Rea­son­ing: $25-50k, 12 weeks

May 19, 2025, 1:25 PM
75 points
1 comment2 min readLW link
(www.flf.org)

Nav­i­gat­ing burnout

gwMay 3, 2025, 10:07 PM
73 points
1 comment9 min readLW link
(www.georgeyw.com)

$500 + $500 Bounty Prob­lem: Does An (Ap­prox­i­mately) Deter­minis­tic Max­i­mal Re­dund Always Ex­ist?

May 6, 2025, 11:05 PM
72 points
11 comments3 min readLW link

Claude 4

Zach Stein-PerlmanMay 22, 2025, 5:00 PM
71 points
24 comments1 min readLW link
(www.anthropic.com)

Bet­ter Air Purifiers

jefftkMay 11, 2025, 4:50 PM
71 points
21 comments3 min readLW link
(www.jefftk.com)

New score­card eval­u­at­ing AI com­pa­nies on safety

Zach Stein-PerlmanMay 26, 2025, 4:00 PM
71 points
6 comments1 min readLW link

What’s go­ing on with AI progress and trends? (As of 5/​2025)

ryan_greenblattMay 2, 2025, 7:00 PM
70 points
7 comments8 min readLW link

Nega­tive Re­sults on Group SAEs

Josh EngelsMay 6, 2025, 9:49 PM
70 points
3 comments8 min readLW link

An­thropic is Quietly Backpedal­ling on its Safety Commitments

garrisonMay 23, 2025, 2:26 AM
70 points
7 commentsLW link
(www.obsolete.pub)