Iván Arcuschin

🚨Wanna know how to increase reasoning behaviors in thinking LLMs? Read our recent work! 👇

Can frontier models hide secret information and reasoning in their outputs? We find early signs of steganographic capabilities in current frontier models, including Claude, GPT, and Gemini. 🧵

Excited to share our paper: "Chain-of-Thought Is Not Explainability"! We unpack a critical misconception in AI: models explaining their Chain-of-Thought (CoT) steps aren't necessarily revealing their true reasoning. Spoiler: transparency of CoT can be an illusion. (1/9) 🧵

With @Butanium_ and @NeelNanda5 we've just published a post on model diffing that extends our previous paper. Rather than trying to reverse-engineer the full fine-tuned model, model diffing focuses on understanding what makes it different from its base model internally.

AI Control is a promising approach for mitigating misalignment risks, but will it be widely adopted? The answer depends on cost. Our new paper introduces the Control Tax—how much does it cost to run the control protocols? (1/8) 🧵

🚀 Excited to announce the launch of the AISAR Scholarship, a new initiative to promote AI Safety research in Argentina! 🇦🇷 Together with Agustín Martinez Suñé, we've created this program to support both Argentine established researchers and emerging talent, encouraging…

Lots of progress in mech interp (MI) lately! But how can we measure when new mech interp methods yield real improvements over prior work? We propose 😎 𝗠𝗜𝗕: a Mechanistic Interpretability Benchmark!

New paper w/@jkminder & @NeelNanda5! What do chat LLMs learn in finetuning? Anthropic introduced a tool for this: crosscoders, an SAE variant. We find key limitations of crosscoders & fix them with BatchTopK crosscoders This finds interpretable and causal chat-only features! 🧵