MIT
UC Berkeley
Jacob HiltonVP of Research · Alignment Research CenterResearcher↗
Researcher · Founder · Policy
Paul Christiano
Alignment Research CenterExecutive Director
OpenAI FoundationBoard Member; Safety and Security Committee Member
Christiano was a principal architect of reinforcement learning from human feedback, which trains models using human preferences and became central to modern AI assistants. He also developed influential ideas for scalable oversight and third-party testing of frontier-model risks.
Career & education
Current rolesAlignment Research CenterJul 2026-presentExecutive DirectorOpenAI FoundationSep 2026-presentBoard Member; Safety and Security Committee Member
Alignment Research CenterJul 2026-presentOpenAI FoundationSep 2026-presentPreviouslyAlignment Research Center2021-2024Founder, Executive Director
OpenAI2017-2021Member of Technical Staff
Center for AI Standards and InnovationApr 2024-Sep 2025
Head of SafetyAlignment Research Center2021-2024OpenAI2017-2021Education
Work and education history is primarily focused on AI-relevant roles and may not be comprehensive.
Citation Trend
Total citations to scholarly work, including research papers, technical reports, books and theses, as recorded by Google Scholar. Each point is a dated snapshot.
citations63,689As of Oct 27, '26
citations+10,315Since last snapshot, Jun 15, '26
Updates 2
UpdatePaul Christiano joins the OpenAI Foundation Board
The AI alignment researcher will also serve on the Foundation's Safety and Security Committee and observe the OpenAI Group PBC board.
September 9, 2026MovePaul Christiano returns to lead the Alignment Research CenterChristiano has returned to ARC as executive director, making its research agenda his main focus while continuing limited government advisory work.
August 4, 2026Selected contributions
Training Language Models to Follow Instructions with Human Feedback
With Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang and others ·InstructGPT uses human feedback to train language models to follow instructions.
Deep reinforcement learning from human preferences
Taught AI agents tasks using people's comparisons of their behavior rather than hand-written rewards.
Concrete Problems in AI Safety
Outlined five practical research problems for preventing accidents in AI systems.
Based on public sources. If accuracy is important, verify details using the links or references.
For corrections, updates, or requests to remove information, contact us.
Reviewed September 12, 2026 · Checked for accuracy and new information.
