Paul Christiano joins OpenAI Foundation Board

OpenAI appointed Paul Christiano to the OpenAI Foundation Board and its Safety and Security Committee, chaired by Zico Kolter. He will sit as a non-voting observer on the OpenAI Group PBC Board. The SSC governs safety and security practices across OpenAI, including the for-profit.

Christiano founded the Alignment Research Center, led alignment work at OpenAI from 2017 to 2021—including foundational RLHF research—and has served as a Senior Tech Advisor at NIST's Center for AI Standards and Innovation, working on frontier evaluations with national-security implications. As advisor he will recuse from OpenAI-related matters and model evaluations.

Board chair Bret Taylor framed the pick as strengthening oversight with someone rigorous on the hardest capability questions. Christiano said rapid capability gains make the SSC's job more important and more challenging, and that alignment remains a difficult technical problem.

We're treating this as a governance signal with real weight and real limits. Christiano is not a decorative safety hire. He helped define modern alignment practice, left to build independent research capacity, then worked inside the U.S. standards apparatus on frontier evaluation. Putting that profile on the Foundation Board and SSC says the nonprofit control layer wants someone willing to argue that catastrophic risk is a live design constraint.

The structure matters as much as the résumé. Foundation Board and SSC membership is where OpenAI says safety authority lives after the 2025 recapitalization and attorney-general reviews. Non-voting observer status on the PBC board means influence without a formal for-profit vote—voice in the room, not a veto on the cap table.

Useful AI still needs people who will slow a ship date when evidence is thin. Christiano's public posture has long taken seriously that advanced systems could pose catastrophic risk and that safeguards should precede dangerous capability levels. If the SSC actually gates deployments on evidence, this appointment strengthens that muscle. If the committee becomes a briefing theater while product orgs set the pace, the hire becomes reputational cover.

I'm watching whether SSC decisions show up as delayed launches, narrowed agent permissions, or published evaluation bars—and whether Christiano's independence survives proximity to the lab he once helped build.

OpenAI says the appointment builds on governance commitments tied to California and Delaware attorney-general reviews around the October 2025 recapitalization. The Foundation is a separate nonprofit that controls OpenAI Group PBC and holds a significant equity stake while running its own charitable work.