Jacques Thibodeau

Montreal, Canada

Bio

Updated 03/22/26

Jacques Thibodeau is a French Canadian AI alignment researcher and physicist working to reduce risks from superintelligent AGI. He is the founder of a stealth AI safety organization focused on building infrastructure to progressively automate technical AI safety research, helping researchers rapidly iterate on experiments through a centralized alignment research ecosystem. He created the Alignment Research Literature Dataset, which has been used by OpenAI, Anthropic, and StampyAI, and has conducted research on model editing techniques and mechanistic interpretability. He participated in SERI MATS in Berkeley (July 2022) and has been an AI Safety Camp participant, including on a project to build datasets and tools for alignment research. He collaborates with Quintin Pope on the Supervising AIs Improving AIs agenda, which focuses on making automated AI science safe and controllable. Prior to his AI safety work, he had a background in physics and government roles in social innovation, strategic foresight, and regulation.