Christian Schroeder de Witt is the Principal Investigator of Oxford Witt Lab (OWL) at the Department of Engineering Science, University of Oxford. He was awarded a Royal Academy of Engineering Research Fellowship jointly with a Schmidt AI 2050 Early Career Fellowship in summer 2025, raising over £3.5m as sole PI within that season. Spanning a diverse background in artificial intelligence, physics, computer science, and climate policy, Christian’s interests range from foundational and theoretical AI research to real-world applications. Most recently, Christian defined the field of multi-agent security, which is addressing a crucial gap in the existing AI security and safety literature by focusing on worst-case guarantees in agentic systems. Within multi-agent security, Christian proposed undetectable threats, a conceptual limit object and autonomous systems security primitive that imposes hard constraints on anomaly-detection based AI assurance and motivates security-by-design approaches. Recently, Christian has illustrated undetectable threats through his work on secret collusion (NeurIPS 2024), illusory attacks (ICLR 2024 Spotlight), and unelicitable backdoors (NeurIPS 2024) conducted with his student Sumeet Motwani, and collaborators Tim Franzmeyer and Andis Draguns (respectively). Christian leads a variety of research collaborations ranging from game theory to long-horizon LLM planning, and open-source intelligence, including with external partners LLNL, BBC Verify, Adobe, and Google DeepMind and he has published not just in top-tier AI venues, but also security venues such as AsiaCCS and IEEE WIFS.

Prior to founding OWL, Christian’s work (with Sam Sokota) solved a 25-year old problem in information security, perfectly secure steganography, named a 2023 Biggest Discovery in Computer Science by Quanta Magazine. During his PhD, Christian helped pioneer the field of deep multi-agent reinforcement learning alongside Jakob Foerster. As a Computer Science master’s student at Oxford working with Bob Coecke, Christian proved an open incompleteness theorem for categorical quantum mechanics. His MPhys thesis on complex systems/theoretical biological evolution with Ard Louis was awarded an Oxford university thesis prize.