OpenAI's new math advisory board has nine members and zero actual power
OpenAI has created an Advisory Group on Mathematics and AI, based at Princeton's Institute for Advanced Study, to evaluate and help coordinate the publication of its AI-generated math results. The group has nine members and no authority over OpenAI's research pace or direction — a limitation both OpenAI and the IAS have stated explicitly. That detail matters now because the math community is already pushing back hard on how the company handles its claimed breakthroughs.
The backdrop
The advisory group follows OpenAI's announcement that its internal model solved the Navier-Stokes equation — one of mathematics' seven Millennium Prize problems — and tackled more than 100 other open problems on top of that. The pace of those announcements triggered a backlash: on September 11, 25 Fields Medalists (the highest honor in mathematics) signed an open letter criticizing how AI labs are racing to solve famous problems. Their concern is that this frenzied publication style is damaging mathematics' intellectual culture before the broader community can properly verify results.
The Navier-Stokes claim itself is already under scrutiny. As Scientific American reports, experts including University of Chicago mathematician Luis Silvestre argue the model exploited an artificial external-force formulation — potentially solving a different, easier version of the problem than the one the Millennium Prize targets.
A symbolic gesture
The advisory group's stated job is to assess the significance of OpenAI's mathematical results and coordinate how they're published. Members can speak publicly, shape the group's own composition, and offer recommendations. They won't be paid.
What they cannot do is pump the brakes. Of the nine mathematicians on the board, only one — Camillo De Lellis — signed the Fields Medalists' critical letter, per TechCrunch. The other 24 signatories have no seat at the table. That gap between the board's composition and the breadth of academic concern is hard to ignore.
This fits a pattern of OpenAI's AI models drawing scrutiny for behavior that outpaces oversight — from data practices to, now, how breakthroughs get announced. The company has also been expanding aggressively on the hardware side, recently partnering with Samsung to develop its own AI chips.
What changes
In practice, probably little. OpenAI keeps control of its research timeline. The advisory group offers a veneer of academic accountability without the mechanisms to enforce it. Whether that's enough to quiet the math community's concerns seems unlikely — the Fields Medalists who signed that letter weren't asking for a board. They were asking for a slower, more careful process. This board can't deliver that.