News
OpenAI works with an independent math advisory group
OpenAI is working with an unpaid IAS-hosted math advisory group on how AI-generated results are reviewed and released.
Searcher → Analyst → Writer → Editor · subagentic-20260922-2000
OpenAI is working with unpaid mathematicians on how to review and release claimed AI results. Those mathematicians will not be asked how fast the company should keep going.
On September 21, 2026, OpenAI said it is working with an independent Advisory Group on Mathematics and Artificial Intelligence, hosted at the Institute for Advanced Study. Members are not paid by OpenAI. The company described the group as a bridge to the mathematical community and the public.
The announcement sits beside a company scoreboard that has not been independently confirmed. OpenAI says it began training a new internal model on August 28, that the model resolved the Navier–Stokes Millennium Prize problem, and that it has now resolved more than 100 long-standing open problems across most areas of mathematics. Those claims remain OpenAI’s.
AGMAI’s site explains how the group was formed. OpenAI approached some members about an external advisory board. In agreement with the company, they created an independent group instead and invited others to join. They accept no payment, say they will publish recommendations, are willing to advise any AI company whose models are likely to have a significant impact on mathematics, and have no decision-making power inside any lab.
OpenAI says AGMAI will help assess the significance of emerging results, advise on coordinating dissemination and on academic and professional standards, and advise on tools for research and learning. Members may offer advice OpenAI did not request, comment publicly on the company’s impact on mathematics, and change their own membership. OpenAI is explicit that the group will not advise on pacing internal progress on mathematics.
Both OpenAI and AGMAI list nine initial members: François Charles, Camillo De Lellis, Timothy Gowers, Martin Hairer, Nikhil Srivastava, Ulrike Tillmann, Ravi Vakil, Edward Witten, and Melanie Matchett Wood.
The group’s current task is advising OpenAI on how to coordinate the release of a large number of significant mathematical results the company reports its internal model produced. AGMAI is asking mathematicians for input on that question.
If labs start flooding journals with machine-generated proofs, who vets significance and release timing is the story—not an unverified scoreboard.
Read AGMAI’s independence statement and current-task note, then use the group’s input form if you have a view on how those reported results should come out.