OpenAI appointed AI-safety researcher Paul Christiano to the OpenAI Foundation board on Sept. 9, placing a prominent alignment specialist inside the nonprofit body that controls the company’s for-profit arm. Christiano will also join the Foundation board’s Safety and Security Committee and serve as a nonvoting observer on the OpenAI Group PBC board.
The appointment brings an experienced internal critic closer to OpenAI’s model-release and risk-governance process as the company pushes its newest systems into the market. For executives adopting frontier AI, the move is a reminder that vendor governance is becoming part of operational due diligence—not a concern that can be separated from product performance.
A new role with two lines of oversight
OpenAI said the Safety and Security Committee oversees safety and security practices across the organization, including OpenAI Group PBC. The distinction in Christiano’s roles matters: He is joining the Foundation board, while his seat on the operating company’s board is observational and does not carry a vote.
Christiano is a senior technical adviser to the U.S. Center for AI Standards and Innovation, part of the National Institute of Standards and Technology. OpenAI said he will recuse himself from government work involving the company and from OpenAI model evaluations. He also founded the nonprofit Alignment Research Center.
Technical experience meets governance responsibility
From 2017 to 2021, Christiano led alignment research at OpenAI and contributed to foundational work on reinforcement learning from human feedback, a training approach widely used in modern AI systems. His career has focused on whether increasingly capable models remain aligned with human intentions and under meaningful human control.
TechCrunch and Axios independently confirmed the appointment. OpenAI’s official announcement said the selection builds on a governance structure created during the company’s October 2025 recapitalization.
Why the board appointment matters
AI governance is moving from broad policy language into specific questions of authority: Who can challenge a release, what evidence they can demand and where final accountability sits. Christiano’s addition strengthens the technical safety expertise available to the committee, but the practical test will be how that expertise shapes decisions when safety concerns collide with commercial speed.
Customers should not treat the appointment as proof that every risk is resolved. It is better read as a governance signal. Procurement, legal and technology leaders evaluating OpenAI should ask how the company’s safety committee exercises oversight, how incidents are disclosed and what continuity plans exist if a model’s release or access changes. Those questions now belong in enterprise AI strategy alongside cost, security and capability.