Submissions from lesswrong.com

1.		Quantity vs. Quality in Pottery Is a Lie (lesswrong.com)
		1 point by barry-cotter 1 day ago \| past \| 1 comment
2.		Cat-Belling Problems (lesswrong.com)
		1 point by surprisetalk 1 day ago \| past \| discuss
3.		Quantity vs. quality in pottery is a lie – LessWrong (lesswrong.com)
		4 points by amanverasia 2 days ago \| past \| 1 comment
4.		Lessons from Red-Teaming Auto-Review, OpenAI's Blocking Monitor (lesswrong.com)
		1 point by gmays 2 days ago \| past \| discuss
5.		Leaking Muse's Charter, Meta's model spec (lesswrong.com)
		2 points by gmays 3 days ago \| past \| discuss
6.		Against SURA (lesswrong.com)
		2 points by eatitraw 4 days ago \| past \| discuss
7.		Plan R: AI Safety by ASICs (lesswrong.com)
		2 points by neverbeenasked 6 days ago \| past \| discuss
8.		The Ability of Frontier Models to Control Robots Is Rapidly Improving (lesswrong.com)
		1 point by johnsutor 6 days ago \| past \| discuss
9.		Cat-Belling Problems (lesswrong.com)
		4 points by neverbeenasked 7 days ago \| past \| discuss
10.		The Ability of Frontier Models to Control Robots Is Rapidly Improving (lesswrong.com)
		2 points by gmays 8 days ago \| past \| discuss
11.		GLM-5.3 Sometimes Considers How Train-Time Actions Shape Future Behavior (lesswrong.com)
		2 points by eatitraw 9 days ago \| past \| discuss
12.		Fermi Poker (lesswrong.com)
		2 points by danielfetz 11 days ago \| past \| 1 comment
13.		How could the US and China verifiably slow AI? (lesswrong.com)
		1 point by gmays 11 days ago \| past \| discuss
14.		What's the Date? (lesswrong.com)
		3 points by surprisetalk 13 days ago \| past \| discuss
15.		[flagged] I Did This Without Your Knowledge (lesswrong.com)
		75 points by jmugan 13 days ago \| past \| 48 comments
16.		Twitch Plays: Interpretability of Bonsai-8B (lesswrong.com)
		1 point by mordymoop 13 days ago \| past \| 1 comment
17.		How does Fable 5.1 hide messages in images? (lesswrong.com)
		219 points by gmays 13 days ago \| past \| 74 comments
18.		I did this without your knowledge (lesswrong.com)
		1 point by amanverasia 14 days ago \| past
19.		Arithmetic is an underrated world-modeling technology (lesswrong.com)
		1 point by amanverasia 14 days ago \| past
20.		GLM 5.3 internally considers its responses influence on next model training (lesswrong.com)
		1 point by vld_chk 14 days ago \| past
21.		Defense Against the Super-Worms (2025) (lesswrong.com)
		1 point by 7777777phil 14 days ago \| past
22.		A Critical Look at Lean4 (lesswrong.com)
		1 point by paulpauper 14 days ago \| past
23.		GLM-5.3 Sometimes Considers How Train-Time Actions Shape Future Behavior (lesswrong.com)
		4 points by Gathering6678 15 days ago \| past
24.		I Did This Without Your Knowledge (lesswrong.com)
		3 points by -_- 16 days ago \| past
25.		Neural networks are not already clean (lesswrong.com)
		3 points by signa11 17 days ago \| past \| 1 comment
26.		Measuring Intelligence Beyond Human Scale (lesswrong.com)
		3 points by nicholascarolan 17 days ago \| past
27.		Neural networks are not already clean (lesswrong.com)
		3 points by signa11 18 days ago \| past \| 1 comment
28.		Language models revise the documents that govern them (lesswrong.com)
		1 point by paulpauper 20 days ago \| past
29.		Cat-Belling Problems (lesswrong.com)
		1 point by amanverasia 21 days ago \| past
30.		Some interesting internal documents from Anthropic litigation (lesswrong.com)
		6 points by gosen 21 days ago \| past \| 1 comment
		More