In a post on X, @OpenAI says it is working with an independent advisory group of mathematicians to help it responsibly share advances in AI and mathematics. OpenAI’s September 21 announcement presents the group as a bridge to the mathematics community and broader public—not as a regulator, proof-checking authority, or body that controls the company’s internal research schedule.

The announcement comes after OpenAI reported that a new internal model had resolved more than 100 long-standing mathematical problems, including the Navier–Stokes Millennium Prize problem.

What OpenAI’s mathematics advisory group is

OpenAI presents the group as giving mathematicians a voice in how it reviews and communicates emerging mathematical results and develops tools related to mathematical research and learning. Its stated role covers four broad areas:

  • reviewing emerging mathematical results and assessing their significance;

  • communicating and coordinating the dissemination of those results;

  • advising on academic and professional standards for mathematical research; and

  • advising on tools that support mathematical research and learning.

The group is hosted at the Institute for Advanced Study. OpenAI says its broader goal is to put capable tools in mathematicians’ hands so they can pursue questions in their fields and develop new ideas. The announcement separately says the group will not advise the company on how to pace its internal mathematical progress.

Why OpenAI says it created the group

OpenAI says the rapid progress of its internal mathematical systems prompted discussions about how to inform the wider community and help the field prepare and adapt. The company also points to concerns raised in a recent open letter, “A Severe Misalignment of AI in Mathematics,” about the negative externalities of using solutions to open problems as benchmarks for AI systems.

In that context, OpenAI presents the advisory group as a way to engage with mathematicians as AI systems become more capable. The announcement frames responsible development and deployment as important beyond mathematics because mathematical discoveries may lead to a wide range of applications.

What the group will advise on

The group’s stated remit covers the significance and communication of mathematical results, research standards, and tools for mathematical research and learning. OpenAI says members will help assess the significance of new results and advise on how to coordinate their dissemination.

That dissemination role matters because an AI-generated result can require expert review and context before it is useful to researchers or the public. The company also says the group will advise on academic and professional standards, linking the announcement to questions about how AI-assisted mathematical work should be evaluated and presented.

A separate part of the remit concerns research and learning tools. OpenAI says the group will advise on how its tools can support mathematicians and learners, but the announcement does not describe a specific product, evaluation process, or tool recommendation.

How OpenAI says the group will remain independent

OpenAI describes several provisions intended to let the group challenge the company rather than simply endorse its work. According to the announcement, members will be able to:

  • offer advice that OpenAI did not request;

  • comment publicly on OpenAI’s impact on mathematics;

  • make their advice public; and

  • change the group’s membership as they see fit.

OpenAI also says its members will not be paid by the company. The company presents members’ ability to exercise their own judgment and challenge OpenAI’s decisions as central to the group’s value.

These are provisions OpenAI says will govern the group’s independence. The supplied announcement does not provide further details about selection criteria, terms of service, meeting procedures, conflicts of interest, or a schedule for publishing advice.

What the group will not control

The announcement draws one explicit boundary: the group will not advise OpenAI on how to pace its internal mathematical progress. Its stated remit therefore does not include advising OpenAI on how quickly to pursue its internal mathematical progress.

The group can advise on the significance, communication, standards, and research and learning tools described in the announcement, while OpenAI retains responsibility for its internal research decisions under the structure the company describes.

The nine initial members

OpenAI lists these initial members of the Advisory Group on Mathematics and Artificial Intelligence:

  • 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 announcement provides these affiliations but does not include biographies or explain how the initial members were selected. Listing them does not establish that they have endorsed OpenAI’s reported mathematical results.

What the announcement does—and does not—establish about OpenAI’s results

The advisory group was announced alongside OpenAI’s account of rapid progress from a new internal model. In a separate September 8 article, the company said its system produced an analytical proof and a formalization of the argument in Lean, a proof assistant—a system for encoding mathematical arguments in a form that can be checked. OpenAI said the formalization was verified. The result concerned the Navier–Stokes existence and smoothness problem.

The Navier–Stokes equations describe fluid motion. The associated Millennium Prize problem concerns three-dimensional, incompressible fluid flow with constant density and asks whether a solution that starts smoothly can develop a singularity in finite time. Here, a singularity means that fluid speeds grow without bound within a finite time under the conditions described by the problem.

OpenAI says its system produced a proof of one of the problem’s official alternatives and that Lean formalization and verification followed. The company also says it does not intend to claim the Millennium Prize for the result.

Those details help explain why the company says it wants advice on reviewing and communicating emerging mathematical results. They do not show that the new advisory group has reviewed the proof, confirmed it, or accepted responsibility for the result. The announcement also does not provide evidence that the group has already met, issued advice, or changed how OpenAI handles the Navier–Stokes work.

For now, the group is best understood as an independent advisory structure that OpenAI says will help connect its mathematical AI work with mathematicians and the broader public. Its effectiveness will depend on how openly it can publish criticism and advice, how independently it can change its membership, and how those stated powers are used in practice.

Sources