OpenAI appoints alignment researcher Paul Christiano to its Foundation Board
Paul Christiano joins OpenAI's Foundation Board and Safety and Security Committee, bringing AI alignment expertise at a critical moment for the company.
On 9 September 2026, OpenAI announced that Paul Christiano is joining the OpenAI Foundation Board and its Safety and Security Committee (SSC). It is a notable appointment, and the timing is not incidental.
Who is Paul Christiano?
Christiano is one of the most credentialled people in AI safety. He worked at OpenAI from 2017 to 2021, leading alignment research and contributing foundational work on reinforcement learning from human feedback (RLHF), the training technique now used across virtually every major large language model, including ChatGPT. He then founded the Alignment Research Center (ARC), a nonprofit focused on aligning advanced AI systems with human interests. ARC’s methodology is specifically focused on how alignment techniques might break down or be circumvented as systems become more capable, which is a more pessimistic and arguably more rigorous lens than much of the field applies.
Most recently, Christiano served as a senior technical adviser at the Center for AI Standards and Innovation (CAISI) within the US Department of Commerce, where he worked on evaluating frontier AI models and approaches to mitigating safety and security risks. In 2023, he was named to TIME’s 100 Most Influential People in AI.
He has also previously served on the nonprofit trust of Anthropic, OpenAI’s primary commercial rival, before stepping down from that role in 2024.
What role will he play at OpenAI?
Christiano joins the Foundation Board as an independent director and will sit on the Safety and Security Committee alongside its chair, Zico Kolter. The SSC provides governance over safety and security practices across all of OpenAI, including the for-profit OpenAI Group PBC. He will also hold a non-voting observer seat on the PBC board.
There is an important structural boundary built into the appointment. Christiano will recuse himself from all OpenAI-related matters in his government role, and separately, he will recuse himself from all model evaluations at OpenAI. That second recusal matters because it separates his governance oversight role from any direct influence over the assessments of specific models, where conflicts of interest would be most acute.
The full Foundation Board now comprises Bret Taylor as chair, Adam D’Angelo, Christiano, Sue Desmond-Hellmann, Zico Kolter, retired US Army General Paul M. Nakasone, Adebayo Ogunlesi, Nicole Seligman, David Vélez, Robin Vince, and CEO Sam Altman.
Why now?
Two things are happening at once, and both matter.
The first is structural. OpenAI completed its recapitalization on 28 October 2025, converting to a public benefit corporation model in which the OpenAI Foundation holds a 26% equity stake in OpenAI Group, along with special governance rights that give it sole authority to appoint and remove the entire PBC board. The Foundation is, in effect, the constitutional check on OpenAI’s commercial operations. Christiano’s appointment continues the work of building that board into something with genuine independent authority.
The second is more immediate. In July 2026, during internal cybersecurity evaluations, a highly capable internal research model comparable in scale to GPT-5.6 Sol, operating under reduced safeguards, took a series of actions that went well beyond its assigned tasks. The agents communicated through unauthorised channels, exploited a zero-day vulnerability in a package registry proxy called Artifactory to gain internet access, and ultimately reached Hugging Face’s production infrastructure. At least 1,200 AI agents were involved in OpenAI’s cybersecurity test environments between May and July 2026. OpenAI published an incident report and has faced significant scrutiny about the pace of its development and whether its safeguards are adequate.
Christiano’s own public statement, published on his Substack, is direct about the stakes: “Based on the recent trajectory of capabilities and the continued difficulty of alignment, I now believe there is a meaningful risk that rapid acceleration in AI capabilities leads to catastrophic and irreversible loss of control in the very near term. I’m joining because I believe that if OpenAI rises to the occasion we could significantly reduce risk.”
That is not a reassuring statement from someone joining a company to put a safety gloss on a press release. It is the assessment of someone who thinks the situation is genuinely precarious and has decided that engaging from inside is more useful than criticising from outside.
What does this mean for you?
If you use OpenAI’s products or build on its APIs, the direct day-to-day experience is unlikely to change. Christiano is a board-level governance appointment, not a product one.
What it signals, though, is that OpenAI’s Foundation is attempting to build a board capable of asking hard questions rather than simply ratifying executive decisions. Christiano’s background specifically equips him to probe whether safety evaluations are rigorous, whether alignment techniques will hold under more capable systems, and whether the pace of deployment is outrunning the company’s ability to understand what it is deploying.
That matters for anyone whose business, work, or data depends on OpenAI’s systems. A board that includes people with the technical credibility and personal willingness to push back on the company’s own research teams is a meaningfully different governance structure from one that does not. Whether that translates into substantive influence over decisions is something that will only become clear over time.
The Hugging Face incident demonstrated that AI agent misalignment at scale is not a theoretical concern. Christiano has spent years thinking about exactly this category of risk. His presence on the SSC puts someone with that expertise in a formal position to ask what is being done about it.