OpenAI millennium proof dispute: Mathematician accuses lab of academic fraud

A public dispute over an AI-produced solution to one of mathematics' most famous unsolved problems has escalated into accusations of academic fraud against OpenAI. Mathematician Tristan Buckmaster has accused the lab of misrepresenting his role in work that produced a proof related to the Navier–Stokes existence and smoothness problem, one of the seven Millennium Prize Problems. OpenAI CEO Sam Altman has flatly rejected the allegations, but the clash has spilled into the open and is forcing researchers to ask whether they can trust AI companies to follow scientific norms.

The controversy centers on a 2024 result that used AI to make progress on a problem that has stumped mathematicians for decades. Buckmaster, a Princeton mathematician who specializes in fluid dynamics, says OpenAI improperly credited him as a collaborator or adviser in a way that made it appear he endorsed the work. He has called the lab's actions a form of academic fraud, arguing that the company leveraged his reputation without his consent to lend credibility to an unverified result.

Altman pushed back in a series of public statements, insisting that OpenAI had followed proper protocols and that Buckmaster's characterization was wrong. The CEO did not provide a detailed rebuttal, leaving the mathematical community to parse a he-said-she-said between a prominent academic and one of the most powerful AI companies in the world. The lack of a transparent, third-party investigation has only deepened concern.

The dispute has drawn a sharp warning from Terence Tao, one of the most respected mathematicians alive and a Fields Medal winner. Tao cautioned that cases like this could "reverse centuries of tradition in open science," where results are shared, scrutinized, and credited carefully. If AI labs are seen as willing to bend those conventions, he argues, researchers may stop trusting the scientific output that comes from them—and may refuse to collaborate at all.

For mathematicians, the stakes go beyond this single proof. AI tools are increasingly used to generate conjectures, check proofs, and even propose new approaches to old problems. But those tools are built by companies with commercial incentives, and the norms for sharing data, code, and credit are still being written. Buckmaster's accusation suggests that the standard practice of listing authors or advisers can be exploited to make an AI result look more legitimate than it is.

The immediate next step is unclear. Neither OpenAI nor Buckmaster has called for an independent review, and the mathematical community has no formal mechanism to arbitrate disputes between a researcher and a private lab. Tao's warning implies that the damage may already be done: if researchers cannot trust that their names and contributions are being used fairly, the collaboration between academia and AI labs—which many see as essential for future breakthroughs—could grind to a halt.

The broader question is whether AI labs will adopt the transparency norms that have governed science for centuries, or whether they will continue to operate under their own rules. The answer will determine not just who gets credit for the next big result, but whether the scientific community can rely on AI-generated work at all.