METR

@METR_Evals57k followers

We work to scientifically measure whether and when AI systems might threaten catastrophic harm to society. Nonprofit.

Open on X

Top outlier

18.5×

Best organic post

Peak views

64k

Single post

Bookmarks

415

Across 2 organic posts

Last scan

Oct 16

Community cache

Posts 2 posts

#	Date	Post	↓ Outlier	Views	Likes	Reposts	Bookmarks
1	Oct 14	If an AI developer was on track to lose control of its internal agents, would anyone outside the company know? Today, we are sharing METR’s vision for how we would work with AI dev	18.5×	64k	647	109	251
2	Oct 6	In order to notice when AI agents misbehave, AI companies often log the actions and reasoning steps their agents take. However, misaligned AI agents may be able to hack the softwar	11.7×	41k	606	61	164
3	Oct 14	In a companion piece, we explain our approach to working with AI developers in exercises like these. Our goal is to get the public accurate, timely, and independent evidence so tha	2.6×	9k	76	2	11
4	Oct 14	(3) AI R&D acceleration If AI automates AI R&D, capability growth could outpace what alignment and control methods can handle. Our proposal is to analyze how much AI automa	2.2×	7.5k	81	3	14
5	Oct 14	(1) Alignment The public should know if agents would attempt actions like those seen in incidents like the hack of Hugging Face. Our proposal is to analyze a developer’s internal i	1.3×	4.4k	82	7	13
6	Oct 14	We describe three critical areas where AI developers should provide better open evidence: (1) alignment, (2) control, and (3) AI capability acceleration. In each area, we detail th	1.0×	3.5k	80	5	8
7	Oct 14	Our proposed exercises would provide evidence, not confident safety assurance or formal oversight. We also encourage developers to work with other external groups to complement the	0.7×	2.4k	55	0	3
8	Oct 6	Treating agent observability as security-critical infrastructure means handling all AI outputs (transcripts, reasoning, actions, etc.) as untrusted inputs: ideally we can make it v	0.7×	2.3k	54	0	1
9	Oct 14	(2) Control Agents that attempt harmful or unwanted actions are not always noticed. Our proposal is to analyze whether a developer’s internal agents could get away with serious har	0.7×	2.3k	69	2	6
10	Oct 6	View post on X	0.6×	2.3k	42	1	6
11	Oct 14	This proposal is our recommendation for how we could give the public visibility into risks occurring within AI development. We stand ready to conduct all three of our proposed exer	0.4×	1.6k	54	1	3

Scan another creator

Run this on any account.

XOutlier builds this table for every public profile you open on x.com, with live data and every post, not just the top 50.

Add to Chrome Download .zip