Overview
- OpenAI announced early this week that one of its advanced, nonpublic models produced a theoretical proof and a simulation showing a fluid singularity for the Navier‑Stokes existence and regularity question.
- The company says it reached the result by running large numbers of autonomous software agents, first using about 1,000 agents on a simplified Euler variant and then scaling to roughly 10,000 agents and millions of dollars in compute to find the demonstration after about 88 hours.
- Two mathematicians who have been working on the problem, Tristan Buckmaster and Levent Alpöge, say OpenAI began work only after their progress became known and they suspect the company’s models may have used their unpublished drafts, an allegation OpenAI denies.
- The Clay Mathematics Institute continues to classify the Navier‑Stokes problem as unsolved because the work has not yet been published or independently reviewed, and OpenAI has said it would decline the $1 million prize if the proof is confirmed.
- Experts warn that AI systems that recombine published and private research could reduce transparency and discourage researchers from sharing partial results, and they note that journals and the math community will need full, reproducible documentation to validate any AI‑derived proof.