Particle.news

OpenAI Says Its Model Solved the Navier–Stokes Problem as Mathematicians Demand Credit and Review

OpenAI reports a model produced a proof after an 88‑hour, large‑agent run, prompting calls for clearer attribution, fuller citation, and independent mathematical validation

Overview

  • OpenAI announced its internal study describes a model-generated solution to the Navier–Stokes Millennium Problem after running thousands of agents in a coordinated 88‑hour campaign.
  • The company told researchers it will not claim the Clay Millennium Prize and acknowledged Tristan Buckmaster and Levent Alpöge’s prior role in moving the work forward.
  • Spanish researchers Diego Córdoba and Luis Martínez‑Zoroa say their 2023 vorticity‑cascade method supplied the core idea the model used, and they and others say early OpenAI drafts omitted important citations.
  • Buckmaster has publicly accused OpenAI of pressing him over publication terms and of limited transparency, while OpenAI denies using identifiable user data but allows anonymized derivative data could have helped train models.
  • Leading mathematicians and organizations, including the RSME and 25 Fields Medal winners, are demanding rigorous, independent proof checking and new norms on credit, citation and how companies use vast compute in academic discovery.