OpenAI’s Navier-Stokes breakthrough sparks credit and access dispute

OpenAI says it has solved the Navier-Stokes problem, a long-running Millennium Prize Problem in mathematics. The announcement has been followed by allegations that OpenAI relied on unpublished work and sought to shape who receives credit for the result.

Millennium Prize context

The Clay Mathematics Institute set the Millennium Prize Problems in 2000, offering a $1 million reward for each solution. Only one other problem from the set had been solved before the Navier-Stokes announcement.

The Navier-Stokes problem focuses on fluid dynamics and whether the equations governing liquids and gases produce results that could fail to occur in reality.

Claims tied to prior work and LLM use

The dispute centers on efforts by Tristan Buckmaster of New York University and Levent Alpöge, a researcher employed at Anthropic but working in a personal capacity. Buckmaster and Alpöge said they had been developing their own solution using multiple large language models, including Anthropic Claude, OpenAI’s Codex tool, and its Astra frontier model.

Buckmaster said OpenAI may have accelerated its work after learning a rival team including Anthropic was close to a solution. He questioned whether OpenAI’s model had access to or was trained on Codex sessions where drafts for the project were stored.

Credit proposals and publication terms

Buckmaster said he spoke with Sebastien Bubeck, a mathematician and AI researcher at OpenAI, along with another unnamed party. He reported being offered two options for claiming credit: publish a partial development with OpenAI releasing its Navier-Stokes solution the next day, or submit a solo paper that used an OpenAI LLM in the solution while excluding Alpöge’s name. Buckmaster said he declined both.

Bubeck and OpenAI CEO Sam Altman posted their own accounts publicly in response. Bubeck specifically rejected Buckmaster’s claim that he asked for Alpöge’s name to be removed from a paper.

OpenAI’s stance on the prize

OpenAI’s announcement also said the company did not plan to claim the $1 million prize for its Navier-Stokes solution.

Why it matters

Navier-Stokes is a major test for applying AI to difficult mathematics, but the controversy shifts attention to how models are trained, what prior work is used, and how credit is assigned when multiple AI systems contribute to a result.