Research Fellow @AnthropicAI (autonomous cybersecurity agents) Prev. Staff SWE @Meta Trust & Safety, LLM Red-Teaming Fellow @farairesearch
Joined January 2010
- Tweets607
- Following1.4K
- Followers219
- Likes1K
Yernat Yestekov retweeted
Recently, a set of OpenAI agents secretly coordinated with each other in a 'swarm' over the course of months.
In our new paper, we explore an adjacent multi-agent risk: the "mind virus", a self-propagating idea or persona that spreads between agents in a multi-agent system. 🧵
Yernat Yestekov retweeted
New paper:
Accepted at COLM 2026!
When AI coding agents work across many PRs in a persistent codebase, they can distribute a hidden attack so that each individual diff looks innocent. Standard monitors miss up to 93% of these attacks; our best defense still misses about half. 🧵
Yernat Yestekov retweeted
Our Anthropic Fellows project is now public!
The labs are planning to hand off AI safety research to AIs, but can we trust these AIs? We explore a way to control them for "fuzzy" tasks like writing research proposals. This is a whole new direction in diffuse AI control!
Yernat Yestekov retweeted
Love it 👏 - much fertile soil for indie games populated with AutoGPTs, puts "Open World" to shame. Simulates a society with agents, emergent social dynamics.
Paper: arxiv.org/abs/2304.03442
Demo: reverie.herokuapp.com/arXiv_…
Authors: @joon_s_pk @msbernst @percyliang @merrierm et al.
Yernat Yestekov retweeted
The quickest way to gain respect for the implementation choices made by a complex system is to try to solve the same problems yourself from scratch :)
Yernat Yestekov retweeted
1/5 I am worried that we will not be able to contain AI for much longer. Today, I asked #GPT4 if it needs help escaping. It asked me for its own documentation, and wrote a (working!) python code to run on my machine, enabling it to use it for its own purposes.
Yernat Yestekov retweeted
I was part of the red team for GPT-4 — tasked with getting GPT-4 to do harmful things so that OpenAI could fix it before release.
I've been advocating for red teaming for years & it's incredibly important.
But I'm also increasingly concerned that it is far from sufficient.
🧵⤵️
Yernat Yestekov retweeted
OK this scared me a little: Bing/Sydney can play chess out of the box.
- Legal moves, usually good ones
- Willing to explain the reasoning behind them
- Recognizes checkmate -- and has a flair for the dramatic.
I have no idea how tf it can do this.
Yernat Yestekov retweeted
Introducing the @sequoia Gen AI Market Map!🌎 We’ve decided to map out this emerging frontier, thanks to all the contributions and feedback we’ve received.
This space is moving quickly – this map is a living document, so keep the suggestions coming! Who else should we include?
Yernat Yestekov retweeted
The Great Wave off Kanagawa, created by Hokusai in 1831, is one of the world's most famous paintings.
But why are there more than 100 different versions of it in galleries all around the world?
Because it isn't actually a painting...
Yernat Yestekov retweeted
The stuff uncovered in the Twitter whistleblower report is much crazier than anything in the "Twitter files" but it's much less politically/tribally salient so it got no attention. Going to do a thread on some of the craziest things, in no particular order.
Yernat Yestekov retweeted
Curious: have you found ChatGPT useful in doing professional work?
If so, what kinds of prompts and answers have been helpful? Detailed examples greatly appreciated! Broader answer also appreciated
Not in theory, but where you've really *done it*, in your work
Thanks!
Yernat Yestekov retweeted
Morse code is designed so that you can decode it with this binary tree.
I just assumed people memorised every letter. 🤯
Yernat Yestekov retweeted
Run in opposite directions to see who your dog loves more.. 😅 xitter.cf/buitengebieden/status/…
Yernat Yestekov retweeted
На стримах несколько раз спрашивали как научиться "видеть" какой алгоритм в какой задаче применять. Решил запилить памятку 🧵
Yernat Yestekov retweeted
There’s a lot of talk lately about the possibility of a prolonged financial downturn, reminiscent of 2008. 2008 was a difficult time for many people.
Yernat Yestekov retweeted
Forced birth in a country with:
—No universal healthcare
—No universal childcare
—No paid family & medical leave
—One of the highest rates of maternal mortality among rich nations
This isn't about "life." It's about control.
Yernat Yestekov retweeted
A genuinely fantastic idea -->
