Tom David is Cofounder and CEO of the GPAI Policy Lab, a French organisation fostering international cooperation on the security and control of general-purpose AI, with research and strategic analysis addressing the loss-of-control and safety risks posed by frontier models, the verification mechanisms to assess them, and the levers that can enable international coordination. He is an appointed expert to the EU AI Scientific Panel. Tom has worked on general-purpose AI security since GPT-2, with particular attention to CBRN risks and loss of control. He co-founded PRISM Eval, an initiative dedicated to the safety evaluation of frontier general-purpose AI systems. His work has been recognised by the French Defence Innovation Agency and COMCYBER. He advises the IFIP programme committee on AI for Global Security and was a Research Scholar of the ML Alignment Theory Scholars programme in Berkeley. Tom previously worked as a policy officer at Institut Montaigne, with earlier experience at the French Gendarmerie Research Centre and in the Cabinet of the Prefect, where he worked on crisis management and security policy. He has founded several organisations and contributed to a range of AI safety research projects.
Tom holds a degree from the École Normale Supérieure (ENS) de Lyon.
