Not here anymore - at the other place.
Joined March 2013
- Tweets 4,475
- Following 1,091
- Followers 2,529
- Likes 13,661
232 Photos and videos
Ayo - I'm finally off here. Keeping the account to retain my bookmarks, but no longer active. Join me here.
bsky.app/profile/agstrait.bs…
3
519
AS retweeted
Our evaluations show that frontier AI's cyber capabilities are advancing quickly. The length of cyber tasks frontier models can complete has been doubling every few months, and this rate has become faster over time, with recent models exceeding our previous trends. 🧵
31
125
589
137,418
AS retweeted
We know AI systems occasionally act against their operators’ intentions – but what in their environment causes them to do so?
In a new paper, we make progress on this question 🧵
12
24
104
13,677
AS retweeted
Apr 23
We (@AISecurityInst) tested GPT-5.5 for its cyber capabilities and safeguards. It's the strongest performing model we've tested on our narrow cyber tasks and solved one of our cyber ranges in 1/10 attempts. We found a universal jailbreak with 6 hours of expert red teaming.
17
54
371
51,363
AS retweeted
Mar 18
Disturbing anecdotal reports of "AI psychosis" and negative psychological effects have been emerging in the news. But what actually happens during these lengthy delusional "spirals"? In our preprint, we analyze chat logs from 19 users who experienced severe psychological harm🧵👇
24
82
400
53,095
AS retweeted
4 Dec 2025
Did you know that one base model is responsible for 94% of model-tagged NSFW AI videos on CivitAI?
This new paper studies how a small number of models power the non-consensual AI video deepfake ecosystem and why their developers could have predicted and mitigated this.
1
11
40
8,008
🤝 You’ll work with 2 other researchers and in collaboration with other gov departments. The first project is to create a problem book of methods to reduce these risks (building on papers.ssrn.com/sol3/papers.…). Crucially, you are not expected to view sensitive material directly.
26
AS retweeted
In a new paper, we examine recent claims that AI systems have been observed ‘scheming’, or making strategic attempts to mislead humans. We argue that to test these claims properly, more rigorous methods are needed.
4
25
84
17,204
AS retweeted
10 Jun 2025
Newest @reboot_hq 🎙️post: @jessicadai_ and I discuss forecasting, and how people present unhelpful narratives about the future (mostly by picking on AI 2027, sorry guys)
Why we should view the future as constructed, not predicted
4
11
56
4,133
AS retweeted
7 Jun 2025
Apple just GaryMarcus'd LLM reasoning ability
217
560
4,916
3,494,064
AS retweeted
12 May 2025
Advanced AI systems require complex evaluations to measure abilities, but conventional analysis techniques often fall short.
Introducing HiBayES: a flexible, robust statistical modelling framework that accounts for the nuances & hierarchical structure of advanced evaluations.
ALT An example of hierarchically nested evaluation data.
2
11
53
7,273
AS retweeted
15 Apr 2025
How will AI impact the economy? Can we defend against misuse? What policies would mitigate the risks of AI?
Thrilled to share that @random_walker and I are writing another book to tackle these questions! Today, we release a paper laying out our argument: AI as Normal Technology.
ALT Image of the first page of the paper, available at https://kfai-documents.s3.amazonaws.com/documents/c3cac5a2a7/AI-as-Normal-Technology---Narayanan---Kapoor.pdf
12
69
282
58,411
I too find this really weird, mainly in that it shows the frontier of AI research is at risk of moving further away from producing useful, safe, reliable products. These seem like features, not bugs.
2
262
AS retweeted
3 Apr 2025
We've funded 20 new research projects to enhance AI security in critical infrastructure ⚡
Our Systemic AI Safety Grants Programme, announced at the Seoul AI Summit, has awarded up to £200,000 seed grants to projects tackling AI risks 🧵👇
1
10
59
9,537
A great thread re: problematic extrapolations on claims about AI being superhuman at tasks.
1. Coding =/= all computer-related tasks, let alone all tasks
2. Generating code to complete a task =/= the most efficient, secure way to complete a task.
24 Mar 2025
This tweet is misleading. State-of-the-art AI models struggle at some tasks that take humans <10 minutes, while *simultaneously* excelling at some tasks that would take humans several hours or days to solve.
1
299
I'll do a longer post about the new role, but in short, we're building a research team, funding programme, and partnerships to tackle crucial questions about advanced AI's societal impacts. We'll track how AI is being used across critical sectors, and study societal-level risks.
1
1
36
I'm so proud of Ada's accomplishments - from pioneering work on COVID-tech, deep dives into AI auditing and research ethics, supporting emerging AI policy and regulation, exploring risks of AI & genomic systems, foundation models and personal AI assistants...it's a very long list
1
44
But I'm even more proud of the people we've worked with, the community we've fostered, and the culture we've built. I remember the first meeting I had with @carlykind_ about her vision for the org - I was all in. Looking back, we've accomplished so much more than we'd thought.
33