Paul Christiano

Paul Christiano is a prominent figure in artificial intelligence safety, known for his contributions to AI alignment, the discipline focused on ensuring AI systems act in accordance with human values and intentions. His research centers on AI alignment and steering AI systems toward human interests. Christiano earned a PhD in statistical learning theory from UC Berkeley, with a doctoral thesis titled "Manipulation-resistant online learning."

At OpenAI, Christiano co-authored the paper "Deep Reinforcement Learning from Human Preferences" and played a key role in the development of reinforcement learning from human feedback (RLHF), which The New York Times described as "a notable step forward in AI safety research." His subsequent work, including "AI Safety via Debate," addressed scalable oversight—supervising AI systems in domains where humans may have difficulty judging output quality. Christiano led alignment research at OpenAI and contributed foundational work on RLHF.

Christiano founded the Alignment Research Center, a nonprofit research institute in Berkeley, California, dedicated to aligning advanced artificial intelligence with human values and priorities. He also launched an initiative to conduct third-party evaluations of frontier models, now housed at Model Evaluation and Threat Research. The Alignment Research Center develops scalable methods for training AI systems to behave honestly and helpfully, with a focus on how alignment techniques might break down or be circumvented as systems advance.

Christiano has government experience through his work at the Center for AI Standards and Innovation within the National Institute of Standards and Technology, where he has contributed to evaluating frontier AI models, including those with national security implications, and to approaches for mitigating safety and security risks. Christiano was named to the TIME 100 Most Influential People in AI and appointed to the UK government's Frontier AI Taskforce advisory board. He also serves as an initial trustee on Anthropic's Long-Term Benefit Trust.

OpenAI announced Christiano's appointment to the OpenAI Foundation Board, where he serves as a non-voting observer on the board of OpenAI Group PBC and joins the Safety and Security Committee of the Foundation Board, which provides governance over safety and security practices across OpenAI. Christiano's career spans foundational research, nonprofit leadership, and government advisory roles, establishing him as a leading voice on the future of safe and beneficial artificial intelligence.