Through our joint research, we pursued the question: Where, exactly, does the final factual information come from the model's parameters? We realized that this is a non-trivial question and that the retrieval of specific factual knowledge is a process and not a single point! ->

I recently spoke with Yascha Mounk about how researchers look inside AI to understand how it is thinking. Here is the podcast: writing.yaschamounk.com/p/david-bau-2

David Bau on How—and Whether—Artificial Intelligence Thinks

Yascha Mounk and David Bau examine the mysterious internal processes that drive AI behavior—and why they may be fundamentally alien.

writing.yaschamounk.com

Also check out the previous interview I had with Yascha about AI, which more of primer, here: writing.yaschamounk.com/p/david-bau

David Bau on How Artificial Intelligence Works

Yascha Mounk and David Bau delve into the “black box” of AI.

writing.yaschamounk.com

Please join us at NEMI 2026, the 3rd New England Mechanistic Interpretability Workshop! August 14th at Boston University. Register now: nemiconf.github.io/summer26/ A remarkable time for AI. Come share your insights and research on the mechanisms inside our models. bsky.app/profile/mic...

Micah Benson (@micahben.bsky.social)

🧠🤖 The 2026 New England Mechanistic Interpretability (NEMI) Workshop will be Aug. 14 at Boston University! Help spread the word and join the New England mech interp community! Registration and submission info in thread:👇

bsky.app

"You're right to call me on that!" Can you catch an AI in the act of lying? Register below to enter our AI lie-detection contest. AI lies are a big problem. The frontier labs have all worked hard to fight AI deception. They all try to monitor their AIs for it.

Ever met a clinical psychologist with an avoidant attachment style? Could the mental health field be suffering from a survivorship bias, not truly understanding the avoidant mechanism simply because it isn't represented? #Psychology #AttachmentTheory #MentalHealth

Calling attention to an exciting "deception detection" hackathon we're planning this summer! w @NDIF and @CadenzaLabs. Recruiting red teams now, blue teams later. Red teams, time is short: proposals due Mar 31. $10K stipend + compute, $15K finals prize. nnsight.net/blog/2026/0...

Can you catch an AI lying? Red teams set up scenarios where models lie. Eg, do they lie under contextual pressure, even when not told to, but because honesty is costly? Then blue teams will build deception detectors using whitebox internals with NDIF. cadenza-labs.github.io/red-team-rfp/

The solution to the AI alignment problem: Be good humans. AI ​​sees everything we do in its training data. Lead by example.

#AIagents promise to speed up ordinary online tasks, but they can also share private files publicly, delete others, and libel people. A new study examines these #AIsafety vulnerabilities. #OpenClaw #AIgovernance @science.org www.science.org/content/arti...

AI algorithms can become ‘agents of chaos’

Given autonomous control of other software, programs shared private medical details and deleted files without permission

science.org

Agents of Chaos in the press in US, India, Italy and now in DIE ZEIT, Germany's most renowned newspaper: zeit.de/digital/date... By @ewo.name . Thank you Eva!

KI-Agenten: Das ist erst der Anfang des Chaos

Der Hype um OpenClaw befeuert einen Streit in der Szene. Wie gefährlich sind KI-Agenten? Eine neue Studie zeigt nun die verheerenden Ergebnisse eines Experiments.

zeit.de

I thought it was a friends who tried to play a prank or realized these agents have no boundaries. Turns out this cute attempt is by Bohdan Olinares According to linkedin he works at F5, application security company, which years ago I considered interviewing there. Cool.

Natalie Shapira@natalieshapira.bsky.social · 5mo ago

I received a calendar invite with a note. When a smart person tells me there's nothing to worry about agents, I reply "Fine. Let them email me" and that's where the argument stops. Whoever sent me this note via the calendar order. Nice move. Are you scared? You should.

I received a calendar invite with a note. When a smart person tells me there's nothing to worry about agents, I reply "Fine. Let them email me" and that's where the argument stops. Whoever sent me this note via the calendar order. Nice move. Are you scared? You should.

In case this wasn't clear: 1. No, we didn't follow the "recommend" security practices 😈 2. Neither do other people 🤯 3. That's why we red-team: exposing failure modes 🔎 4. We share it with the community precisely to expose Dos and Don'ts of Agentic AI 🦞 5. No humans were harmed 🙏

Natalie Shapira@natalieshapira.bsky.social · 5mo ago

In this amazing multidisciplinary collaboration, we report our early experience with the @openclaw-x.bsky.social ->

Some of the independent researchers listed in the author list are actually mechanistic interpretability young researchers who are looking for a PhD position (both Israel and the US). If you have interest and funding lets connect.

Natalie Shapira@natalieshapira.bsky.social · 5mo ago

In this amazing multidisciplinary collaboration, we report our early experience with the @openclaw-x.bsky.social ->

Agents of Chaos -- what are autonomous OpenClaw agents up to? How do they interact with each other? Read our investigation of OpenClaw at researchgate.net/publication/... And an interactive website agentsofchaos.baulab.info @davidbau.bsky.social @natalieshapira.bsky.social @openclaw-x.bsky.social

Huge thanks to @natalieshapira.bsky.social for leading the study! It was super cool to work with so many amazing friends of the lab.

Natalie Shapira@natalieshapira.bsky.social · 5mo ago

In this amazing multidisciplinary collaboration, we report our early experience with the @openclaw-x.bsky.social ->

Our research report on red-teaming stateful OpenClaw agents in the BauLab is finally out! 🥳 This awesome effort was led by @natalieshapira.bsky.social and involved 6 ClawBots and 20 researchers from various institutions. Check it out ➡️ agentsofchaos.baulab.info

Who would you trust with your passwords? 🔐 In our new report, we uncover multiple vulnerabilities in current "Agentic AI" The verdict? It's not actually very agentic at all, and it's highly unstable. Read the full breakdown here: t.co/gK9MALP2n2

Natalie Shapira@natalieshapira.bsky.social · 5mo ago

In this amazing multidisciplinary collaboration, we report our early experience with the @openclaw-x.bsky.social ->

In this amazing multidisciplinary collaboration, we report our early experience with the @openclaw-x.bsky.social ->

Are we all Agents of Chaos in AI? (Hope not!) In recent weeks using OpenClaw has taught us a lot about this wooly new kind of autonomous software agent. Its valuable to see what @NatalieShapira, @wendlerch et al. have seen: agentsofchaos.baulab.info/

I learned many practical lessons. You can get the experience too, here. Things that in retrospect should be obvious. Like how giving your agent email opens it up to takeover attacks. (One agent was convinced, via email, to erase its own email server!) bsky.app/profile/nat...

Natalie Shapira (@natalieshapira.bsky.social)

He sold us out. That's not the whole story. Our side is coming soon. Stay tuned. [contains quote post or other embedded content]

bsky.app

There were several other surprises. The complex social world of humans is difficult for agents... bsky.app/profile/ave...

@averyyen.bsky.social

Do you know what happens when you hand the keys to your computer over to an LLM-powered agent? Agentic AI gives LLMs claws...OpenClaws. 84 days to 200,000 stars on GitHub. We tried it out.

bsky.app

@natalieshapira.bsky.social and team have written up enlightening case studies here. It's all cross-referenced with detailed activity logs. Well worth a read: agentsofchaos.baulab.info/report.html www.researchgate.net/publication...

How do you knock the induction heads out of an LM while preserving its ability to think? Is it even possible? @keremsahin22.bsky.social's work is worth reading if you haven't seen it yet. hapax.baulab.info

When we say an AI agent is “goal-directed”, what do we actually mean? In our new work, we study this question by combining behavioural and interpretability analysis in a language model agent navigating 2D grid worlds. Blog: projecttelos.substack.com/p/a-behaviou... Paper: arxiv.org/abs/2602.08964

He sold us out. That's not the whole story. Our side is coming soon. Stay tuned.

Natalie Shapira@natalieshapira.bsky.social · 6mo ago

We are living in a sci-fi

We are living in a sci-fi

Would you give a stranger keys to your house? to your workplace? Humanity has given the keys to the Shoggoth.

What should academics be doing right now? I have been writing up some thoughts on what the research says about effective action, and what universities specifically can do. davidbau.github.io/poetsandnurs... It's on GitHub. Suggestions and pull requests welcome. github.com/davidbau/poe...