Millennium Prize rumor mix-up sparks OpenAI sprint controversy

After OpenAI announced it had solved a Millennium-class math problem, dramatic backstory emerged on Twitter: a rumor had been circulating that Anthropic had cracked a Millennium Prize Problem (widely speculated to be the Navier-Stokes equations); upon learning this, OpenAI poured millions of dollars in compute and internal models into about a week of intensive work and rushed out an announcement—yet Anthropic had not actually solved the problem. This has sparked disputes over motives and scientific ethics.

Confirmed

• OpenAI announced it had solved a Millennium-class math problem, widely speculated to be the Navier-Stokes equations.
• Edouard Harris disclosed (flagged by him as joking in tone): weeks ago, a rumor did circulate that Anthropic had solved a Millennium math problem, and this rumor prompted OpenAI to attempt the same problem.
• Anthropic researcher Raj Ammanabrolu said this was the first time he was genuinely blown away by an AI result, and noted there is another currently confidential result.

Not yet confirmed

• Whether the rumor originated from or was tolerated by Anthropic employees remains an unverified claim in the posts.
• Details in zivravid's Hebrew-language tweets, such as "an NYU mathematician was persuaded to abandon a collaboration with Anthropic," are labeled as unverified gossip.
• Comparative claims that OpenAI "reportedly made greater progress than Anthropic" lack first-hand sources.

Why it matters

• Critics including reinforcement learning scholar Csaba Szepesvari argued the release was motivated not by empowering mathematicians or solving the problem, but by beating Anthropic to the punch—reflecting how competition among top AI labs is shaping research pace and publication culture.
• If the confidential result Raj Ammanabrolu hinted at is real, it may herald an even more significant capability breakthrough, drawing intense community attention.

2026-09-09 ~ 2026-09-10 · 6 related posts

Full story(20 episodes)→

• Episode 1: Terence Tao Proposes Keeping Some Open Math Problems Off-Limits to AI(2026-09-03, 2 posts)
• Episode 2: Rumors Claim Claude Solved Navier-Stokes; Terence Tao Responds(2026-09-05, 9 posts)
• Episode 3: Rumor: Anthropic Has Solved Navier-Stokes with Claude(2026-09-06, 2 posts)
• Episode 4: Terry Tao slams closed labs' theorem-proving as viral marketing(2026-09-06, 2 posts)
• Episode 5: Claude Did Not Solve Navier-Stokes, Clarifies Terence Tao(2026-09-06, 2 posts)
• Episode 6: AI proof of ζ(5)(3) irrationality sparks debate over what counts as research(2026-09-07, 20 posts)
• Episode 7: Mathematicians Accuse OpenAI's Astra Math Results of Research Misconduct(2026-09-07, 2 posts)
• Episode 8: Terence Tao's latest AI remarks spark buzz, Gary Marcus calls them 'fire'(2026-09-08, 2 posts)
• Episode 9: Rumors swirl that Claude may have solved the Navier-Stokes millennium problem(2026-09-08, 3 posts)
• Episode 10: OpenAI's Navier-Stokes Claim Sparks Authorship and Racing Controversy(2026-09-08, 439 posts)
• Episode 11: Tao Hails AI-Assisted Fluid Equations Blowup Breakthrough, Formalized in Lean(2026-09-08, 8 posts)
• Episode 12: Terence Tao warns AI-driven scooping could erode mathematics and open science(2026-09-08, 16 posts)
• Episode 13: OpenAI Accused of Copying Anthropic Employee's Remarks in Apparent Jab(2026-09-08, 2 posts)
• Episode 14: Gary Marcus Presses Bubeck Over Inflated GPT-5 Math Claims(2026-09-08, 2 posts)
• Episode 15: Chollet: AI Navier-Stokes Proof Only Counts With Novel Insight(2026-09-08, 2 posts)
• Episode 16: AI cracks Navier-Stokes with tens of millions in compute, igniting cost and scoop debates(2026-09-09, 22 posts)
• Episode 17: OpenAI's Navier-Stokes Proof Dispute: Altman and Bubeck Respond to Attribution Claims(2026-09-09, 36 posts)
• Episode 18: Ex-DeepMind researcher casts doubt on disputed AI math claims(2026-09-09, 2 posts)
• Episode 19: AMS Confirms Navier-Stokes Breakthrough as Lean Proof Goes Open Source(2026-09-09, 4 posts)
• Episode 20: AI math breakthroughs demoralize university math community(2026-09-09, 2 posts)

• Anthropic Millennium Prize Rumor Sparks Drama as OpenAI Runs $M Week-Long Sprint to Verify — willdepue · 2026-09-09
• [source] Anthropic researcher says an AI result impressed him for the first time, teases another secret one — iandanforth · 2026-09-09
• [source] Anthropic Millennium Problem rumor spurred OpenAI to make even more progress, claims investor — harris_edouard · 2026-09-09
• [source] OpenAI's math breakthrough sparks drama: NYU mathematician reportedly pressured to drop Anthropic collaborator — ziv_ravid · 2026-09-09
• RL scholar Csaba Szepesvari joins criticism that OpenAI rushed a result to beat Anthropic — CsabaSzepesvari · 2026-09-09
• OpenAI spends $20M of compute to frontrun AI-assisted near-solution of century-old math problem — ctjlewis · 2026-09-10