OpenAI announced today the formation of an independent Advisory Group on Mathematics and Artificial Intelligence, hosted at the Institute for Advanced Study, to advise on the review, significance, and coordinated dissemination of mathematical results produced by its systems, following claims that a new internal model trained since August 28 has resolved the Navier–Stokes Millennium Prize problem and more than 100 additional long-standing open problems.
What Actually Changed
On September 21, 2026, OpenAI published a company announcement stating that on August 28 it began training a new internal model. In addition to resolving the Navier–Stokes Millennium Prize problem (announced September 8), the model “has now resolved more than 100 long-standing open problems across most areas of mathematics.” The company noted that “the pace of its progress in mathematics has surprised the mathematicians within OpenAI,” prompting internal discussions on how to inform the community and prepare the field.
OpenAI referenced the September 11 open letter “A Severe Misalignment of AI in Mathematics” (mathandai.org), in which mathematicians raised concerns about negative externalities of treating open-problem solving as an AI benchmark. The company stated that these criticisms highlight the need for thoughtful engagement and that it is therefore working with mathematicians who have established an independent advisory group.
The group’s stated role is to advise on the review and communication of emerging results: assessing significance, coordinating dissemination, and advising on academic and professional standards of mathematical research. It will also advise on how OpenAI tools can support mathematical research and learning. OpenAI’s stated aim is “to put capable tools in mathematicians’ hands so they can pursue the questions they know best and develop new ideas.”
The group operates independently of OpenAI. Members are unpaid by OpenAI, may offer unsolicited advice, may comment publicly on OpenAI’s impact on mathematics, and may publish their advice. The group controls its own membership. Explicitly excluded from its remit is advising OpenAI on the pacing of internal mathematical progress.
Initial members, as listed by both OpenAI and the group’s site (agmai.org), hosted at the Institute for Advanced Study:
- François Charles (ENS-PSL)
- Camillo De Lellis (IAS, GSSI)
- Timothy Gowers (Collège de France, Cambridge)
- Martin Hairer (EPFL, Imperial College London)
- Nikhil Srivastava (Berkeley, Simons Institute)
- Ulrike Tillmann (Oxford, INI)
- Ravi Vakil (Stanford)
- Edward Witten (IAS)
- Melanie Matchett Wood (Harvard)
The group’s own site states that it came together after OpenAI approached some members about an external advisory board; the members instead created an independent group and invited others. Its current task is advising OpenAI on coordinating the release of the large number of significant results the company reports its internal model has produced. The group welcomes community input via a form on agmai.org and will publish recommendations on that site. It affirms it has no decision-making power at any AI company.
Signal Implications
- Infrastructure & product reality: OpenAI is publicly asserting rapid, broad mathematical capability advances from a single internal model still in training, with results spanning a Millennium Prize problem and >100 additional open problems.
- Talent dynamics: The involvement of Fields Medalists and leading researchers (including Hairer, a signatory of the critical open letter) formalizes a high-status channel between the mathematical community and OpenAI.
- Product continuity / vendor risk: The advisory structure creates an external review and dissemination layer for mathematical claims; release timing and academic standards are now subject to independent advice, though internal research pace remains solely OpenAI’s.
- Rights, regulatory, or diligence surface: The group’s independence, public-comment rights, and publication of advice introduce a visible external accountability mechanism specific to mathematical results.
- Capital & buyer confidence: The move responds directly to documented community concerns about misalignment, signaling an attempt to manage reputation and legitimacy around AI-generated mathematical discovery.
Who Needs to Act / Check
Founders and technical leads whose products or diligence processes rely on OpenAI mathematical claims should track the group’s forthcoming recommendations on agmai.org and any coordinated release process for the claimed results. Communications advisors preparing board or investor materials on AI scientific capability should treat the September 21 announcement and the group’s independence provisions as the current primary reference for OpenAI’s engagement posture with the mathematical community.