OpenAI backs independent math advisory group at IAS after its AI claims 100+ open problems

OpenAI has set up an independent advisory group of mathematicians, hosted at the Institute for Advanced Study (IAS) in Princeton, New Jersey, to weigh in on how the company handles mathematical research. Announced Monday, the move coincided with the company's claim that an internal AI model has resolved more than 100 open problems across most areas of mathematics, according to TechCrunch.

How the group came together

The body, called the Advisory Group on Mathematics and Artificial Intelligence, introduced itself in a guest post on Terence Tao's blog. It said it was formed after OpenAI approached some of its members about establishing an external advisory board; those approached instead agreed with OpenAI to create an independent group and invite others to join.

The group says it will advise AI companies on their interactions with mathematical research and the mathematical community, including how mathematical results are responsibly presented and released. It states that it operates independently of any AI company, that members are unpaid, and that it will publish its recommendations. It also says it is willing to advise any company whose models are likely to have a significant impact on mathematics, not only OpenAI.

Its immediate task

The group's first job is unusually concrete: advising OpenAI on how to coordinate the release of a large number of significant mathematical results that OpenAI reports were produced by its internal model. The group is asking the wider mathematical community for input through a public form.

According to TechCrunch, the announcement follows the abrupt publication of a solution to the Navier–Stokes Millennium Prize problem, and OpenAI says the same internal model behind that result also resolved the 100-plus additional open problems.

The backdrop is tense. Earlier this month, 25 Fields Medal-winning mathematicians signed an open letter arguing that AI labs are threatening their intellectual work as they compete to one-up each other with solutions to famous problems, TechCrunch reports.

Who is on it

Nine mathematicians are named as initial members: 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) and Melanie Matchett Wood (Harvard).

TechCrunch notes that only one member, De Lellis, also signed the Fields Medalists' letter.

Clear limits on its power

For all its independence, the group has no authority over OpenAI's research agenda. Its role is to assess the significance of new results and help coordinate their release. Members can offer unsolicited advice, speak publicly about their views and control the group's own membership, which gives them a measure of separation from the company.

But the group will have no say in slowing down or redirecting OpenAI's mathematical research. According to TechCrunch, OpenAI's post states that the group will not be responsible for advising the company on how to pace its internal progress on mathematics. The IAS made the same point in its own announcement, saying that while it will give advice, it holds no decision-making power at any AI company, and responsibility for a company's decisions rests with that company.

Why it matters

This is one of the most explicit attempts yet to build external governance into AI-driven scientific discovery. If OpenAI's claims hold up, the sheer volume of machine-generated mathematics would strain traditional peer review, and someone has to decide how such results are verified, sequenced and communicated to the field.

The group's design is a live test case: real independence, public recommendations and unpaid, senior membership, but advisory only, with the lab retaining full control over the pace of its own research. How that balance works in practice — and whether the mathematical community accepts coordinated releases of AI-produced proofs — could set a template for other disciplines as AI labs begin generating research results faster than academia can absorb them.