With the large influx of submissions and a faster pace of research, reproducibility is more important than ever. With this reproducibility challenge, we want to put the focus on best practices wrt. baselines🧱, ablations🌈, eval🔎 and generalizability🗺️ of interpretability!

Despite the huge inflow of researchers, much of the work in interpretability remains anecdotal. Our new repro challenge at BlackboxNLP (co-located with EMNLP 2026) aims to attract work challenging common assumptions and showing failure/success cases of popular methods. Negative results welcome!

As a computer scientist, we INVENTED the phrase "artificial intelligence." It was never exclusively yours. In both fact and fiction, AI has always been awesome, and beautiful, and, yes, problematic — and it still is.

Excellent survey on causal interpretability by @amuuueller.bsky.social and many BauLab members, don't miss it!

IMO a key skill of a good scientist is moving comfortably between the specific & general e.g., relentless in understand the nerdy details of the data (including its coding) WHILE holding onto the big picture of the hypotheses being tested with the data

Today we’re releasing EMO, a new mixture-of-experts (MoE) model trained so modular structure emerges directly from data without human-defined priors. EMO can use a small subset of its experts for a given task while keeping near full-model performance. 🧵

New paper w/ UK AISI: Millions of people now use AI to help them write and communicate. In three experiments (14k participants, 3m+ human ratings) we show that AI writing assistance systematically distorts writer personas – their perceived beliefs, personality, and identity. 🧵

🚨 Tom Griffiths has a podcast where he interviews cognitive scientists podcasts.apple.com/ca/podcast/t... This just went to the top of my list.

2026 is a whirlwind year for AI. Underlying it all is the greatest scientific mystery of our age. How does a neural network think? I talked w Oliver Whang in NYTimes Magazine, on how AI interpretability is a tangle of structure waiting to be unraveled: www.nytimes.com/2026/04/15/...

Tech industry mottos have a mixed track record. But we should hold idealists to their ideals. And we should celebrate when they come through. The Mythos non-release is a remarkable moment of conviction. Thoughts: davidbau.com/archives/20... Bravo to Anthropic's "race the top".

Mfw fiddling with probes all day but patching experiments don't pan out

Thank you for having me! Next time in person! 🤗

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...

I truly believe the rapid advances in the mech interp subfield have something real to offer AI ethics researchers: A chance to look beyond the HOW of evals to the WHY, a first pass at a technical solution when we see the opportunity, a new avenue for showing failures that prove models are not gods

🧵New paper: "Lost in Backpropagation: The LM Head is a Gradient Bottleneck" The output layer of LLMs destroys 95-99% of your training signal during backpropagation, and this significantly slows down pretraining 👇

Check out David's NetHack port! "Complexity does not yield to speed. Judgment remains essential. The work of deciding what matters, of seeing what is hidden, of knowing when your own metrics are lying to you: this is the work that remains, and it is the work worth learning."

This is an important project! If you believe alignment faking is true, you should at least entertain the possibility of misalignment faking before drawing your conclusions. Especially true if researchers fishing for misaligned behaviors are the ones running the evals!

BlackboxNLP is back once again at EMNLP'26! Very happy to be part of the team again, and excited for our new reproducibility track! Check it out ⬇️

tired: meta omni-translation to 1600 low-resource languages wired: kagi translate english to mechinterp

I'm calling it DeepSeek's new 1T parameter model (V4)? The style, content, and length of the reasoning are extremely similar.

My contribution to model welfare efforts for today

This morning I happened to hang out around the Harvard med school café and all conversations I overheard were about LLMs med assistants and XAI 🫡

I want to talk about why AI-based mass surveillance is so dangerous, and why I would oppose it no matter which party or president is in office.

🔥Super excited to share our new demo website for 🪄Interpreto! 🖼️It is basically an explanation gallery showcasing attribution and concept-based explanations for classification and generation. 🎮Play with it: for-sight-ai.github.io/interpreto-d... We will keep improving it, so stay tuned!

Great wrap-up for #EVALITA2026! 🔥 Glad to have helped organize this edition and to see many interesting discussions! Great response to our task Cruciverb-IT (with Ciaccio C., @gsarti.com, Dell’Orletta F., @malvinanissim.bsky.social)! Thanks to all co-organizers and @ailc-nlp.bsky.social! #NLProc

Great release from our engineering team! A lot of the major pain points have been addressed, and this is our first step towards supporting interpretability workflows on more realistic scenarios! Check it out!

Those of us who work in AI in the US today should take a moment to think today. Do not get distracted by the circus. Instead, let us pause to think carefully about our freedoms, our rights, and our responsibilities as citizens and professionals. It is a deadly serious moment.

I'm excited to share that this paper was accepted at ICLR 2026! We show that language models encode one of the most basic ingredients of a world model: the ability to distinguish plausible from implausible states. Check out the paper for more details! See you in Rio! Paper: arxiv.org/abs/2507.12553

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/

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

You can now "pip install latentlens" 🔨 It comes with: * pre-computed embeddings for several popular LLMs and VLMs * a txt file with sentences describing WordNet concepts, which we recommend as a standard corpus to get embeddings from * ... Try it out and let us know what we can improve!

Our ICML 2025 workshop on Actionable Interpretability drew massive interest. But the same questions kept coming up: What does "actionable" mean? Is it achievable? How? We're ready to answer. 🧵

A totally normal morning in Cambridge, MA

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

To accompany our paper, we also released a @hf.co space to explore agent trajectories and decoded cognitive maps from our eval outputs! Find it here: huggingface.co/spaces/proje...

Can LLMs figure out who you are from your anonymous posts? From a handful of comments, LLMs can infer where you live, what you do, and your interests; then search for you on the web. New 📄 w/ @SimonLermenAI, @joshua_swans, @AerniMichael, Nicholas Carlini, @florian_tramer 🧵

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

Karpathy: here's a GPT in 5 lines of pure Python. That's all you need for AGI. Frontier labs: Bernie B2.5 implements GLGMHA (Great Latent Goated Maxi-Head Attention), scaling context size to 1 bazillion tokens

Opus 4.6 still tells me "Good Feedback". Phew, my job is safe (for the next ~3 months)

Surely a good omen, thanks Gemini

Had a bit of fun with @anthropic.com Claude artifacts tonight and ended up with two demos that feel pretty useful for language learners: an assistive reader that lets you export/practice new words, and a translation helper that explains your mistakes. Find them here: gsarti.com/langlearn

Does it matter how you prompt an LLM with a persona? Do LLMs respond differently to natural conversation history compared to names and explicit mentions? Go check out our new preprint! 👀

The Art of Wanting. About the question I see as central in AI ethics, interpretability, and safety. Can an AI take responsibility? I do not think so, but *not* because it's not smart enough. davidbau.com/archives/20...

Can models understand each other's reasoning? 🤔 When Model A explains its Chain-of-Thought (CoT) , do Models B, C, and D interpret it the same way? Our new preprint with @davidbau.bsky.social and @csinva.bsky.social explores CoT generalizability 🧵👇 (1/7)

Can you solve this algebra puzzle? 🧩 cb=c, ac=b, ab=? A small transformer can learn to solve problems like this! And since the letters don't have inherent meaning, this lets us study how context alone imparts meaning. Here's what we found:🧵⬇️

It was an honor to be part of this awesome project! Interpreto is a great up-and-coming tool for concept-based interpretability analyses of NLP models, check it out!

New year, new YouTube videos! We are resuming our regular interpretability seminar posts, with a fantastic talk by Deepti Ghadiyaram on interpreting diffusion models. Watch the video: youtu.be/4eqvABPX5rA

All interpretability research is either philosophy (affectionate) or stamp collecting (derogatory)

The NDIF ecosystem is growing! 🚀 nnterp will bridge the gap between fine-grained fiddling with model internals (nnsight) and low-code access to bespoke viz (workbench). Excited to work with @butanium.bsky.social and the @ndif-team.bsky.social to make it a standard in interp research!