One of the most consequential figures in AI safety now sits on the board that controls OpenAI. Paul Christiano — the researcher who led OpenAI’s alignment team from 2017 to 2021, founded the Alignment Research Center, and now serves as a senior technical advisor at the US government’s Center for AI Standards and Innovation — was appointed to the OpenAI Foundation Board on 9 September. He also joins the Foundation Board’s Safety and Security Committee, the body that provides governance over safety practices across all of OpenAI, working alongside its chair Zico Kolter.
🔍 THE BOTTOM LINE
The man who coined much of the field’s vocabulary for things going wrong has taken a seat inside the structure that is supposed to prevent them — with a recusal agreement that shows how carefully both sides are treading.
Who Christiano Is, and Why the Seat Matters
Christiano’s résumé is effectively a map of the alignment field’s short history. At OpenAI from 2017 to 2021, he led the language model alignment team and contributed foundational work on RLHF — the technique that turned raw language models into assistant-shaped products, including the ones OpenAI sells. After leaving, he founded the Alignment Research Center, a nonprofit focused on theoretical alignment and evaluations of frontier systems, and became one of the few safety researchers whose concerns the labs demonstrably absorbed rather than dismissed.
In 2023 he moved into government, joining what became the Center for AI Standards and Innovation within NIST, where he has worked on evaluating frontier AI models — including capabilities with national-security implications — across two US administrations. That dual identity is the point of the appointment. OpenAI’s announcement describes him as “an independent voice on whether the industry’s safeguards are adequate,” and credits him with taking seriously “the possibility that advanced AI could pose catastrophic risks.”
The Foundation is not a symbolic charity. Under the October 2025 recapitalisation structure, the nonprofit Foundation controls OpenAI Group PBC and holds a significant equity stake in it, with the Safety and Security Committee acting as a committee of the Foundation Board. The appointment was made in connection with the governance commitments OpenAI made following reviews by the California and Delaware attorneys general. In other words: this is the oversight layer, and Christiano is now part of it.
The Structural Tension Worth Watching
Three details in the announcement deserve more attention than the press-release framing suggests.
First, Christiano will serve as a non-voting observer on the OpenAI Group PBC Board — the commercial entity. He gets the board seat and the Safety and Security Committee role at the Foundation level, but at the operating company he can watch, not vote.
Second, the recusal. The announcement’s footnote states that as a Senior Technical Advisor at CAISI, “Paul will recuse himself from all OpenAI-related matters as well as all model evaluations.” Read that carefully: the US government’s frontier-model evaluator, who by OpenAI’s own description has evaluated OpenAI’s models, cannot evaluate OpenAI’s models while holding a seat in OpenAI’s governance. The recusal is designed to protect the government’s independence — and it is also an unusually explicit acknowledgement that the two roles point at each other.
Third, the timing. The appointment lands one day after Jacob Coxon’s resignation from Anthropic made the industry’s internal safety debate public, and weeks after Anthropic’s alignment lead put his own probability on catastrophic outcomes on the record. OpenAI’s own post frames the moment candidly: capabilities “have advanced very rapidly in the last year and alignment remains a difficult technical problem, making the Safety and Security Committee’s responsibility more important and more challenging than ever” — that is Christiano himself, in his quoted statement, describing the job he just accepted.
What It Means
For OpenAI, the appointment buys something money can’t: a safety figure with genuine establishment credibility, at the exact moment regulators in two states have signed off on its governance structure and critics keep arguing its nonprofit control is nominal. For the safety community, the reaction is likely to split — some will read it as the right person finally inside the room; others will note that the room has already absorbed several such figures, and the safeguards they were meant to guarantee remain, by Christiano’s own past statements, unproven.
For New Zealand readers, the relevant point is structural. The people governments trust to evaluate frontier systems and the companies building them are a small, overlapping group — the same names move between labs, agencies, and boards. Whether that’s how you build competence or how you capture a regulator is the question this appointment will be asked to answer, one committee meeting at a time.
❓ FAQ
Who is Paul Christiano? A researcher who led OpenAI’s language model alignment team from 2017 to 2021, contributed foundational work on RLHF, founded the Alignment Research Center, and now serves as a senior technical advisor at the Center for AI Standards and Innovation within NIST, the US government’s frontier-model evaluation body.
What role did he take at OpenAI? He joined the OpenAI Foundation Board and its Safety and Security Committee on 9 September 2026, and serves as a non-voting observer on the OpenAI Group PBC Board, per OpenAI’s announcement.
Does his government role conflict with the board seat? OpenAI’s announcement includes a recusal: as a senior technical advisor, Christiano will recuse himself from all OpenAI-related matters and all model evaluations. That protects his CAISI role but also means the government’s evaluator cannot evaluate OpenAI while sitting in its governance.
Why does the Foundation Board matter? Under the recapitalisation completed in October 2025, the nonprofit Foundation controls OpenAI Group PBC and holds a significant equity stake in it. Its Safety and Security Committee oversees safety practices across all of OpenAI, including the commercial entity.
🔍 THE BOTTOM LINE
OpenAI has put one of the field’s most credible safety researchers inside its governance structure, with the recusal lines drawn in a public footnote. Whether that strengthens oversight or demonstrates how thin the wall between evaluator and evaluated has become is the question the next year of Safety and Security Committee decisions will answer.
📰 Sources:
- OpenAI — Paul Christiano joins OpenAI Foundation Board
- Alignment Research Center
- NIST — Center for AI Standards and Innovation
- Wikipedia — Paul Christiano
— CJ Murden, editor of Singularity.Kiwi. Former digital technologies teacher, author of AI-focused books. Writing with a New Zealand focus.
