OpenAI claims a Navier-Stokes solution, and mathematicians push back

reported by 3 outlets· 6 articles · confidence: medium · first seen 2026-09-09 03:10 UTC

What this means for you

Nothing to act on yet: this is a claim about a proof, not a product, and how independent that proof is remains unsettled. If you use Codex on original mathematics, note the dispute: the Guardian reports one mathematician believes his work reached OpenAI's team; OpenAI denies direct access but will not rule out that his usage improved its model.

On 8 September OpenAI claimed its agents had solved Navier-Stokes, a Millennium Prize problem open for nearly a century. The response was an argument, not applause. Twenty-five Fields medallists have signed an open letter warning of 'severe misalignment' between AI companies and mathematics. The Guardian's editorial questions how independent the work is and says credit to human mathematicians was thin; it reports that the mathematician Tristan Buckmaster believes OpenAI's team saw his own Codex work on the problem. OpenAI denies direct access but cannot rule out that his usage improved its model. OpenAI has since announced a panel of mathematicians to advise it.

Key facts

  • ·OpenAI claimed on 8 September 2026 that its AI agents had solved the Navier-Stokes problem, one of the most difficult open challenges in mathematics. source
  • ·Navier-Stokes had gone unsolved for nearly a century before the claim. source
  • ·Twenty-five recipients of the Fields medal signed an open letter warning of a 'severe misalignment' between AI companies and mathematics. source
  • ·The Guardian reports that mathematician Tristan Buckmaster believes work he did on Navier-Stokes using OpenAI's Codex model was seen by the OpenAI team. source
  • ·OpenAI denied directly accessing that material but could not rule out that data from Buckmaster's use of its products 'helped improve our model'. source
  • ·OpenAI announced a new independent panel of mathematicians to advise it and other AI companies on their interactions with mathematical research, reported on 23 September 2026. source

What the sources say

  • MIT Technology Review AI — Places the episode in the longer argument about what machine-made mathematics does to the field.
  • The Verge AI — Reports how the announcement landed among mathematicians, and why it unsettled them.
  • The Guardian AI — Gathers mathematicians' objections to how OpenAI handled and framed its announcement.
  • The Guardian AI — Interview with Colva Roney-Dougal of St Andrews on what the episode means for mathematics.
  • The Guardian AI — Argues the work's independence is unproven and that human mathematicians were insufficiently credited.
  • The Verge AI — Reports the advisory panel OpenAI set up in response to the damage.

Sources

The original reporting. Follow these — they did the work.

← the wire