Michael Chen

Michael Chen is the AI science advisor to the California Governor’s Office of Emergency Services (Cal OES), placed there through the California Council on Science and Technology’s California AI Science Residency — a program that embeds AI experts inside state agencies to advise on frontier AI safety and risk management. He advises Cal OES leadership on frontier AI safety, critical safety incidents, AI cybersecurity and the risks of internal model deployment.

Chen’s work implements parts of SB-53 — not CalCompute. The same law that establishes the CalCompute consortium in Government Code Section 11546.8 also makes Cal OES the state’s frontier AI reporting authority, and it is that second mandate he advises on. Under the Transparency in Frontier Artificial Intelligence Act, the agency must build the mechanism through which frontier developers and members of the public report critical safety incidents, receive confidential summaries of large developers’ catastrophic-risk assessments from internal model use, review what comes in and — beginning January 1, 2027, and annually thereafter — publish an anonymized, aggregated report to the Legislature and the Governor. Frontier developers have 15 days from discovery to report a critical safety incident.

Chen came to the role from METR, the Berkeley nonprofit that evaluates the capabilities of autonomous AI agents, where he worked on AI policy — helping frontier developers design safety policies and providing input on public policy. His work there included third-party evaluation of sabotage risk and collaboration with U.S. government bodies including the National Institute of Standards and Technology. Before METR he was a software engineer at Stripe, working on backend infrastructure and applications built on large language models.

He is a part-time PhD student at the University of Oxford and an affiliate of the Oxford Martin AI Governance Initiative. He coauthored research on AI deception covered by The Guardian and MIT Technology Review, and has been interviewed in Time on AI evaluations and safety.

SB-53 pairs public AI infrastructure with public AI oversight — CalCompute under the Government Operations Agency, transparency and incident reporting under Cal OES. Chen works on the oversight side of the law. He holds no role in CalCompute itself, and appears here because the statute that created it is the same one he is helping put into effect.