An independent Advisory Group on Mathematics and Artificial Intelligence has formed at the Institute for Advanced Study to advise AI companies on their interactions with mathematical research. Announced on September 21, the initiative starts with a concrete task: advising OpenAI on coordinating the release of mathematical results it says its internal model produced.

The group's founding statement says OpenAI approached some members about creating an external advisory board. They instead agreed to establish an independent group and invite additional mathematicians. Its nine initial members include Timothy Gowers, Camillo De Lellis, Martin Hairer, Ulrike Tillmann and Melanie Matchett Wood.

Advice without decision-making power

Members will not accept payment for this work. The group says it will publish recommendations on its website and is willing to advise any AI company whose models are likely to significantly affect mathematics.

Its mandate includes the responsible presentation and release of results, and communication between researchers and industry. But it explicitly has no decision-making power at AI companies; responsibility for corporate decisions remains with those companies.

That distinction limits what the announcement establishes. The group's description of OpenAI's results attributes them to the company, rather than independently certifying the underlying mathematical claims. Creating an advisory body is not itself validation of a proof.

The group is also soliciting input from the mathematical community to inform its recommendations. For researchers, the immediate development is a public channel for influencing how AI-generated findings are communicated, with published advice rather than binding oversight as its stated mechanism.