Overview
- OpenAI said an internal, unreleased model running about 10,000 cooperating agents produced a proposed solution to the Navier–Stokes existence and smoothness problem after roughly 88 hours of search and about 17 hours of Lean formalization.
- The company published a 166‑page paper and Lean proof files but said it will not seek the Clay Millennium Prize and emphasized that human verification is still required.
- NYU mathematician Tristan Buckmaster publicly accused OpenAI of following an approach similar to his unpublished work with Levent Alpöge and questioned whether use of OpenAI tools exposed their drafts; OpenAI denies direct access but acknowledged it “cannot rule out” that de‑identified usage‑derived data influenced its models.
- Twenty‑five Fields Medal winners published a joint declaration warning that commercial incentives to rush headline results conflict with mathematics’ norms for careful verification and attribution, and OpenAI withdrew sponsorship of a Caltech math event after criticism.
- Independent peer review and specialist scrutiny are now under way and will determine whether the Lean formalization and the argument for a finite‑time singularity hold up, with the Clay Institute’s established prize and vetting processes still in place.