OpenAI claims AI-generated solution to Navier–Stokes Millennium Prize Problem; NYU mathematician disputes it

OpenAI announced a claimed AI-generated solution to the Navier–Stokes Millennium Prize Problem, including a formal Lean proof, reportedly produced by an unreleased model; NYU professor Tristan Buckmaster disputes the result and published competing work with…

On 2026-09-08, OpenAI announced what it describes as an AI-generated solution to the Navier–Stokes Millennium Prize Problem, one of the Clay Mathematics Institute's seven Millennium Prize Problems concerning fluid dynamics (existence and smoothness), carrying a $1,000,000 prize since May 24, 2000. The announcement includes a writeup and a machine-checkable formal proof in the Lean theorem prover. Per Simon Willison's coverage, an unreleased model produced the claimed solution; OpenAI's announcement text itself did not identify the model, methodology, or any independent verification. The result is contested: NYU mathematics professor Tristan Buckmaster accused collaborators of skulduggery and rushed out his own competing results with mathematician Levent Alpoge, who works at Anthropic. The competing claims are described in a hastily published PDF. The reports do not conflict on core facts, but the announcement omitted model and verification details that only the secondary coverage supplies, and the solution's acceptance remains disputed.

• OpenAI announced a claimed AI-generated solution to the Navier–Stokes Millennium Prize Problem on 2026-09-08.
• The announcement includes a writeup and a machine-checkable formal proof in the Lean theorem prover.
• The problem is the Navier–Stokes existence and smoothness problem in fluid dynamics, one of seven Clay Millennium Prize Problems with a $1,000,000 prize since May 24, 2000.
• OpenAI's announcement text did not name the model, methodology, or independent verification; secondary reporting attributes the work to an unreleased model.
• NYU mathematics professor Tristan Buckmaster accused collaborators of skulduggery and published his own competing results with mathematician Levent Alpoge, who works at Anthropic.
• The competing claims are documented in a hastily published PDF; no resolution of the dispute is reported.