OpenAI forms math advisory group after AI solves 100 open problems

By Splitfeed EditorialLast updated Sep 26, 2026, 20:18 UTC

On September 21, 2026, OpenAI announced the formation of an Advisory Group on Mathematics and Artificial Intelligence, to be housed at Princeton's Institute for Advanced Study. The announcement came alongside a claim that an internal model, which began training on August 28, had resolved more than 100 long-standing open mathematical problems within approximately 24 days. Among the problems reportedly addressed is the Navier-Stokes Millennium Prize problem, one of the most famous unsolved questions in mathematics. The group is composed of unpaid mathematicians tasked with advising on how results are assessed, communicated, and released, though TechCrunch reported the panel holds no authority to slow down or redirect OpenAI's ongoing mathematical research.

The stakes here go beyond any single solved theorem. Open problems in mathematics often represent decades or centuries of accumulated effort by the global research community, and the speed and scale of these claims, over 100 problems in under a month, challenge how mathematical knowledge is traditionally produced, verified, and credited. The advisory group's lack of binding power is a specific and concrete point of contention, as is the question of whether any of the claimed results have been independently verified. The structure of the group, unpaid, advisory-only, and announced without prior consultation with much of the mathematics community, shapes how its credibility is being judged.

The pro-AI camp treats the announced results as confirmation of transformative progress, with accounts such as pfdietz suggesting criticism reflects professional defensiveness, and moralestapia dismissing the concerns of mathematicians as the complaints of a group rendered obsolete. Ethan Mollick drew a parallel to other licensed professions, framing mathematics as one more domain that AI is systematically entering. The anti-AI camp is more pointed: Zubiqo described the group as nine unpaid mathematicians with zero power and called it pure PR cover, while cantalopes raised the question of uncredited prior work in connection with the Navier-Stokes result specifically, and bmitc expressed broader frustration with the pattern of Silicon Valley actors setting terms for professional communities without their meaningful input. The middle ground camp acknowledges that some form of outside review is a reasonable step but remains uncertain about its practical effect, with kingstnap questioning how the group would actually manage the dissemination of results, and Education Briefly noting that decision-making remains firmly with the company.

The most immediate questions to watch are whether independent mathematicians verify the claimed solutions, particularly for Navier-Stokes, and whether the advisory group issues any public statements that either affirm or complicate OpenAI's characterizations. Any formal response from prize bodies such as the Clay Mathematics Institute, which administers the Millennium Prizes, would also bear directly on how seriously the broader scientific community accepts these claims.