AI Safety Researcher Paul Christiano Appointed to OpenAI Group PBC Board and Safety and Security Committee

OpenAI appointed pioneering alignment researcher and former federal AI safety official Paul Christiano on Saturday as a non-voting observer on the OpenAI Group Public Benefit Corporation Board and a full voting member of its Safety and Security Committee in San Francisco. The governance appointment brings one of the artificial intelligence field's most respected technical safety architects back into direct oversight of OpenAI's frontier model deployment pipeline. Board directors confirmed that Christiano assumed his committee responsibilities immediately following Saturday morning's governance ratification at the company's 18th Street headquarters.

Christiano, an alumnus of UC Berkeley who previously led OpenAI's early language model alignment team between 2017 and 2021, co-invented Reinforcement Learning from Human Feedback, the foundational training methodology underpinning modern conversational AI systems. After founding the Berkeley-based Alignment Research Center to study autonomous model evaluation and deceptive alignment risks, he served as Head of AI Safety at the U.S. AI Safety Institute within the National Institute of Standards and Technology. His return provides the Safety and Security Committee with deep institutional expertise in pre-deployment red-teaming and national security risk assessment.

Under the company's public benefit governance charter, the Safety and Security Committee holds independent authority to audit frontier training runs, mandate cybersecurity hardening protocols, and delay commercial model releases if preparedness thresholds for biological, cyber, or autonomous replication risks are not satisfied. Technology policy scholars in Berkeley, Lafayette, Orinda, and San Francisco noted that adding a prominent independent alignment theorist strengthens institutional accountability as reasoning models scale into autonomous enterprise and scientific domains.

Academic researchers at UC Berkeley's Center for Human-Compatible AI and East Bay technology ethics boards welcomed the appointment as a constructive bridge between frontier commercial laboratories and the independent alignment research community. Committee officials stated that Christiano will help oversee the upcoming Preparedness Framework 2.0 evaluations, focusing on automated alignment verification, interpretability tooling, and tamper-resistant compute cluster security. Official corporate filings confirm that the Safety and Security Committee conducts mandatory biweekly technical audits with chief safety officers and external government liaisons.

Policy analysts, machine learning researchers, and civic stakeholders can examine the updated Safety and Security Committee charter, preparedness evaluation reports, and board governance disclosures at openai.com or by calling (415) 890-4180. OpenAI and Bay Area academic partners will convene a public symposium on frontier model governance and alignment science in Berkeley on Friday, November 13, at 3:00 PM. Written governance inquiries may be directed to the corporate secretary's office.