Quentin Feuillade--Montixi

Quentin Feuillade--Montixi is a French AI safety researcher and entrepreneur based in Paris. He studied at 42.fr (École 42), a peer-to-peer coding school, and was a scholar in the SERI MATS Winter 2023 cohort (cohort 3), where he was mentored by Janus and Nicholas Kees Dupuis on the cyborgism track and developed expertise in prompt engineering and LLM psychology. After MATS, he worked as a contractor for METR (formerly ARC Eval) evaluating GPT-4 pre-release capabilities, and did independent red teaming for OpenAI and Anthropic. He co-founded PRISM Eval, where he created the BehaviorElicitationTool (BET), an automated red-teaming system that was presented at the Paris AI Summit in 2025. His research interests center on model ethology — a behavioral approach to studying AI cognition — and LLM psychology, topics he has written about on LessWrong and the Alignment Forum. He is currently the founder and CEO of WeaveMind, an AI-powered knowledge management startup backed by Seldon Labs and ElevenLabs startup grants.

No endorsements yet.

Personal Website

https://weavemind.ai/

LinkedIn

LinkedIn

Twitter / X

Twitter

LessWrong

quentin-feuillade-montixi

EA Forum

-