OpenAI announced on Monday an independent advisory group hosted at the Institute for Advanced Study in Princeton, New Jersey, to give mathematicians formal input into the company's math research. The Advisory Group on Mathematics and Artificial Intelligence launches alongside OpenAI's claim that a single internal model has now resolved more than 100 open problems across most areas of mathematics, following the abrupt publication of a solution to the Navier-Stokes Millennium Prize problem.
Nine prominent mathematicians have been named as initial members. Members won't be paid, and they retain the right to offer unsolicited advice, go public with their views, and control their own membership — a structural independence unusual for corporate advisory boards.
“This group will serve as a bridge to the mathematical community and broader public, giving mathematicians a voice in how we move forward”— OpenAI, company blog post
The formation follows an open letter earlier this month signed by 25 Fields Medal-winning mathematicians arguing that AI labs are threatening their intellectual work as they compete to publish solutions to famous math problems. Only one member of the new OpenAI-adjacent group, the IAS's Camillo De Lellis, also signed that letter — a notable gap between the two rosters.
Key facts
- 01OpenAI announced the Advisory Group on Mathematics and Artificial Intelligence on Monday, hosted at the Institute for Advanced Study in Princeton.
- 02OpenAI claims its internal model has resolved more than 100 open problems across most areas of mathematics.
- 03Nine mathematicians were named as initial members; only one, IAS's Camillo De Lellis, also signed the recent Fields Medalists' letter.
- 0425 Fields Medal winners signed an open letter earlier this month warning that AI labs are threatening mathematicians' intellectual work.
- 05The group cannot advise OpenAI on the pace of its internal mathematical research.
The advisory group's remit is narrower than its name suggests. Members will assess the significance of new results and coordinate their release, functioning as an interface between OpenAI's research output and the wider mathematical community. What they cannot do is tell OpenAI to slow down.
OpenAI stated plainly in its post that "the group will not be responsible for advising us on how to pace our internal progress on mathematics." That carve-out addresses the central complaint from the Fields Medalists' letter — the speed at which results are landing — by explicitly refusing to negotiate on it.
The Institute for Advanced Study echoed the boundary in its own release, making clear that hosting the group does not confer authority over any AI company's decisions. The IAS framing preserves the institute's neutrality while acknowledging that the actual choices about model deployment and publication cadence remain OpenAI's alone.
“Although we will give advice, we do not have decision making power at any AI company, and the responsibility for the decisions made by any company will rest with that company”— Institute for Advanced Study, press release
The Navier-Stokes claim is the immediate context. That problem is one of the seven Millennium Prize problems, and any credible solution — from a human team or an AI model — is a landmark event in mathematics. The publication caught the community off guard and intensified concern about how AI-generated proofs should be vetted, credited, and released.
The broader claim of 100-plus resolved open problems raises its own questions about verification workflow. Traditional peer review operates on timescales of months to years per result; an internal model producing results at industrial cadence pressures every part of that pipeline, from refereeing capacity to priority disputes to how credit is assigned when a machine does the work.
The skeptics' case, articulated by the 25 Fields Medalists, is that this cadence risks turning mathematics into a benchmark race between labs — one where the prestige structure of the field is captured by whichever company can publish fastest. The advisory group is OpenAI's answer to that critique, but its lack of power over research pace means the underlying tension is unresolved. Whether nine mathematicians assessing significance after the fact can meaningfully shape norms remains an open question, and the near-total non-overlap with the letter's signatories suggests the harder-line critics are not yet at the table.
For OpenAI, the group is a governance signal at low cost: it buys credibility with the mathematical establishment without ceding operational control, and it puts a respected institution's name on the release process for future results. For the mathematical community, the trade is real advisory input in exchange for accepting that the pace question is off the table. Expect other frontier labs to copy this template — advisory bodies at prestigious institutions, structured to look independent while leaving the core research schedule untouched — as AI systems keep landing on problems that used to define human careers.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
