OpenAI forms an independent Mathematics Advisory Group

OpenAI has established an independent Mathematics Advisory Group of nine mathematicians to guide how the company communicates and develops AI capabilities in mathematics, following an internal model's resolution of more than 100 open problems in the field.

What's new

OpenAI says it is "working with mathematicians who have established an independent mathematics advisory group" to serve as a bridge between the company and the mathematical community. The group is hosted at the Institute for Advanced Study in Princeton and includes nine prominent mathematicians. Members do not accept payment for the work and the group operates independently of OpenAI, meaning it can comment publicly on the company's mathematics work without prior approval.

The advisory group's remit is narrow by design: it is meant to advise on responsible communication and development of AI mathematics capabilities, not to slow or redirect OpenAI's internal research pace. The company has been explicit that the group will not be responsible for advising on how to pace internal progress on mathematics.

The announcement follows OpenAI's disclosure that an internal model, which began training on August 28, has resolved more than 100 long-standing open problems across most areas of mathematics, including a claimed proof of the Navier-Stokes existence and smoothness Millennium Prize problem — a result ModelDex covered when OpenAI first announced it.

Context

The scale of the claimed mathematical results — over 100 open problems from a single model — triggered significant scrutiny and skepticism from the academic mathematics community, including public questions about whether some claimed proofs, such as Navier-Stokes, actually solved the problem as formally stated. Forming an outside advisory board made up of working mathematicians, rather than responding to critiques piecemeal, is OpenAI's structural answer to that scrutiny.

This follows a pattern of OpenAI standing up external advisory bodies around high-stakes or contested capability claims, similar to how other frontier labs have brought in outside experts around safety and alignment work. Unlike a typical corporate advisory board, this one is explicitly positioned as independent enough to criticize OpenAI publicly.

Why it matters

If an AI system really is resolving long-standing open mathematics problems at this pace, verification becomes the bottleneck — and verification is a job for mathematicians, not AI labs marking their own homework. Standing up an independent group with the explicit ability to speak publicly without OpenAI's sign-off is a concession that self-reported capability claims of this magnitude need outside checking to be credible.

It also sets a precedent other labs may face pressure to match: as AI-generated mathematical and scientific claims proliferate, an independent verification layer — funded by but not controlled by the lab making the claims — may become the expected standard rather than the exception.