Submissions from lesswrong.com

		Google Translate apparently vulnerable to prompt injection (lesswrong.com)
		59 points by julkali 2 days ago \| past \| 3 comments
		AI found 12 of 12 OpenSSL zero-days (lesswrong.com)
		2 points by greedo 4 days ago \| past \| 1 comment
		Are We in a Continual Learning Overhang? (lesswrong.com)
		2 points by cubefox 10 days ago \| past \| discuss
		AI found 12 of 12 OpenSSL zero-days (lesswrong.com)
		2 points by jelsisi 10 days ago \| past \| discuss
		A Simple Method for Accelerating Grokking (lesswrong.com)
		2 points by vuciv 11 days ago \| past \| discuss
		Test your interpretability techniques by de-censoring Chinese models (lesswrong.com)
		2 points by allenleee 11 days ago \| past \| discuss
		How AI is learning to think in secret (lesswrong.com)
		2 points by jstanley 12 days ago \| past \| 1 comment
		AI discovers 12 of 12 OpenSSL zero-days (while curl cancelled its bug bounty) (lesswrong.com)
		7 points by excited-dev-11 12 days ago \| past \| 1 comment
		Good if make prior after data instead of before (lesswrong.com)
		13 points by surprisetalk 13 days ago \| past \| 8 comments
		Does Pentagon Pizza Theory Work? (lesswrong.com)
		3 points by nreece 15 days ago \| past
		How AI Is Learning to Think in Secret (lesswrong.com)
		3 points by mannykannot 16 days ago \| past \| 1 comment
		Dangerous capabilities can suddenly appear from gradual progress in AI (lesswrong.com)
		1 point by DalasNoin 18 days ago \| past
		Deep Learning as Program Synthesis (lesswrong.com)
		2 points by todsacerdoti 20 days ago \| past
		Shallow review of technical AI safety (2025) (lesswrong.com)
		1 point by ofou 20 days ago \| past
		Metacompilation (lesswrong.com)
		2 points by Antibabelic 20 days ago \| past
		Evidence that METR may be underestimating LLM time horizons (lesswrong.com)
		1 point by aorobin 21 days ago \| past
		Reflections on TA-ing Harvard's first AI safety course (lesswrong.com)
		2 points by sebg 25 days ago \| past
		Lies, Damned Lies and Proofs: Formal Methods Are Not Slopless (lesswrong.com)
		94 points by OgsyedIE 26 days ago \| past \| 44 comments
		The Exit (lesswrong.com)
		3 points by notarobot123 28 days ago \| past
		AI Teddy Bears: A Brief Investigation (lesswrong.com)
		3 points by surprisetalk 29 days ago \| past
		Humanity Wins (lesswrong.com)
		2 points by PhilosophyForAI 30 days ago \| past
		On Owning Galaxies (lesswrong.com)
		5 points by optimalsolver 32 days ago \| past
		An interactive toy model for exploring AI's effect on the labour market (lesswrong.com)
		1 point by ebursztein 33 days ago \| past
		Opinionated Takes on Meetups Organizing (lesswrong.com)
		2 points by surprisetalk 34 days ago \| past
		Insights into Claude Opus 4.5 from Pokémon (lesswrong.com)
		123 points by surprisetalk 34 days ago \| past \| 23 comments
		Chesterton's Fence (lesswrong.com)
		3 points by foster_nyman 34 days ago \| past
		You Will Be OK (lesswrong.com)
		3 points by walterbell 38 days ago \| past
		Straussian Memes (lesswrong.com)
		43 points by kp1197 39 days ago \| past \| 52 comments
		You Will Be OK (lesswrong.com)
		3 points by sebg 39 days ago \| past
		Eliezer s unteachable methods of sanity (lesswrong.com)
		1 point by prakashqwerty 40 days ago \| past
		More