NYU mathematician Tristan Buckmaster has accused OpenAI of using unfair tactics after his team's progress on a Millennium Prize problem was reportedly leaked to the rival AI lab, which then raced to produce its own proof using vastly greater computing resources. Buckmaster and Anthropic mathematician Levent Alpöge had been quietly working on a specific, uncommon approach to the "Navier-Stokes existence and smoothness" problem before learning that information about their progress had reached OpenAI, which subsequently claimed to have achieved a full proof itself. OpenAI's Sebastian Bubeck has denied the allegations as "false and inflammatory," saying he acted according to academic norms and promising a fuller response.

The dispute centres on one of the seven Clay Mathematics Institute Millennium Prize problems, each carrying a $1 million reward, with the Navier-Stokes equations being widely used but poorly understood theoretically. Buckmaster and Alpöge announced three proofs, including a preliminary finding, using a mix of AI models such as OpenAI's Codex and Claude. Buckmaster alleges OpenAI was evasive about when its research began and how much human input was involved, eventually conceding its first prompt was sent only days earlier, after news of the pair's work had circulated. He further claims Bubeck asked him to remove Alpöge's credit as part of a proposed compromise, a request he refused.