OpenAI keeps bulldozing mathematicians

On September 21, a collection of leading mathematicians announced the formation of the Advisory Group on Mathematics and Artificial Intelligence (AGMAI) in a guest post on the blog of Fields Medalist Terence Tao. The post and OpenAI’s separate statement described AGMAI as an independent body of nine eminent mathematicians intended to advise frontier AI labs on reviewing and communicating emerging mathematical results. OpenAI emphasized the group’s independence while also noting AGMAI would not advise the company on how to pace its internal mathematical progress.

Despite those assurances, several mathematicians and observers learned about AGMAI primarily through OpenAI’s public rollout, which contributed to confusion about whether the group was genuinely independent or effectively appointed by the company. AGMAI’s website acknowledges that OpenAI approached some of the mathematicians who later joined, though it also states the group decided to form independently and invites collaboration with other labs. It remains unclear which members OpenAI initially contacted.

Martin Hairer, an AGMAI member and professor at Imperial College London and EPFL, told The Verge the group does not receive financial or technical support from OpenAI and that members have not signed restrictions beyond standard confidentiality needed for early access to research. Hairer said the announcement and surrounding communications felt rushed and that OpenAI’s wording did little to dispel concerns about potential spin. He described the formation period as intense and frustrating and acknowledged the group has only recently started meeting and planning.

OpenAI’s motivation for convening AGMAI appears tied to a pending release of a large volume of mathematical claims: the company says an unreleased internal model has resolved the Navier–Stokes Millennium Prize problem and “more than 100 long-standing open problems across most areas of mathematics.” That anticipated cascade of results has already alarmed researchers worried about credit, scooping, and how rapid public disclosures might affect the field. AGMAI’s first task — helping coordinate how those findings are reviewed and communicated — places it at the center of efforts to manage relations between mathematicians and AI labs during a fraught period.