Jacob Tsimerman, a 2026 Fields Medal recipient, announced on July 23 in Philadelphia that he will join OpenAI to focus on AI safety1.
Tsimerman's Fields Medal recognized his proof of the André-Oort conjecture. He represented Canada at the International Mathematical Olympiad in 2003 and 2004, winning two gold medals including a perfect score in 2004. He earned his Ph.D. from Princeton in 2011 under advisor Peter Sarnak and became the youngest full professor in the history of the University of Toronto's mathematics department. He is the first mathematician to win the Fields Medal while employed at a Canadian university.
The hire places one of the world's most decorated mathematicians inside OpenAI's safety operation. Boaz Barak tweeted that he was "excited and deeply humbled" to welcome Tsimerman to work on AI safety. Sebastien Bubeck also welcomed Tsimerman to the OpenAI team.
Tsimerman's move carries a particular tension: he has publicly expressed concern that AI has significant potential to cause human extinction. In 2025, he uploaded a paper titled "A Taxonomy of Existential Futures Involving AI" with Andrew Critch, which classifies and gives examples of extinction-level events caused by AI. He has said he supports pausing AI development but does not believe a pause will happen. He said it is better to get involved with AI and help shape it than resist from the outside.
Despite his safety concerns, Tsimerman is an active user of AI tools. He said he uses ChatGPT and Claude to define problem boundaries, locate references, and have models prove lemmas. He said AI has doubled his research efficiency. He also said the proofs produced by AI are usually incorrect but often help him identify the right approach. He said AI may surpass mathematicians within two years at proving theorems. He uploaded five mathematics papers to arXiv in 2025.
Tsimerman's move to industry mirrors a broader pattern documented in the sources. Over the past year, Google DeepMind, OpenAI, and Anthropic have competed intensely for mathematical talent. Tsimerman's former student Levent Alpöge, a graduate of Harvard's Society of Fellows whose undergraduate thesis advisor was Tsimerman, is now a mathematician at Anthropic. Alpöge recently posted on X that the Jacobian conjecture — proposed in 1939 and unsolved for 87 years — is false. Claude Fable 5 helped Alpöge find the counterexample, which was independently verified. Timothy Gowers said this was the first time he had seen an LLM solve a problem famous in the field outside his own area.
ANALYSIS Tsimerman's hiring signals that OpenAI is investing elite mathematical talent specifically in its safety function rather than solely in capability research. His public record of existential-risk concern, combined with his stated rationale for joining — shaping AI from the inside rather than resisting from outside — positions the hire as a credibility marker for OpenAI's safety commitments at a time when the lab faces scrutiny on that front.
The teacher-student split — Tsimerman at OpenAI, Alpöge at Anthropic — illustrates how the competition for top mathematical minds now extends to researchers whose primary identity is in pure mathematics rather than machine learning.
Tsimerman was born in 1988 in Kazan, Russia, and moved with his family first to Israel and then to Canada. He entered the University of Toronto at age 16 and completed his undergraduate studies in two years.