FULL STORY
The Navier-Stokes Rumor and Terence Tao's Responses
Terence Tao proposed shielding open math problems from AI, then a rumor spread that Anthropic had solved the Navier-Stokes Millennium Problem. Tao repeatedly pushed back, calling it viral marketing and clarifying the claim was unsupported.
2026-09-03 ~ 2026-09-06 · 4 episodes · 15 posts
Episode 1 · Terence Tao Proposes Keeping Some Open Math Problems Off-Limits to AI (2026-09-03, 2 posts)
Terence Tao argues that pre-AI open math problems are scarce uncontaminated benchmarks, and proposes designating some as off-limits to AI solvers since solving them carries irreversible costs.
- Terence Tao proposes keeping some math problems off-limits to AI solvers — Outside-Iron-8242 · 2026-09-03
- Terence Tao: Some math problems may need to be off-limits to AI solvers to preserve their value — NathanpmYoung · 2026-09-04
Episode 2 · Rumors Claim Claude Solved Navier-Stokes Millennium Problem; Terence Tao Responds (2026-09-05, 9 posts)
Starting September 5, rumors spread on X that Anthropic had Claude solve one of the Millennium Prize Problems—the existence and smoothness of the Navier–Stokes equations—with the paper reportedly under expert review; nothing has been officially confirmed to date.
Confirmed
- Source of the rumor: Andrew Curran predicted Anthropic had solved the problem and bet the company would announce it before its IPO; mathematician Elliot Glazer reposted approvingly, and his past predictions about frontier AI math ability have been quite accurate (m3, m7)
- Mathematician Julia Stadlmann uploaded a related 34-page paper to arXiv, one of the sparks of the discussion (m9)
- Terence Tao himself responded: clarifying that he was not involved and unaware, and denying that his earlier posts hinted an AI lab had cracked the problem—a guess robinhouston had made (m4, m9)
- Tao took the opportunity to post a long piece voicing concerns about AI math ability: AI solves problems too fast with a black-box process, potentially "polluting" the research value of major open problems in mathematics. Using global regularity of Navier–Stokes as an example, he stressed that historically "failed" attempts gave rise to important theories such as Leray–Hopf weak solutions—the process matters more than the answer (m8, m9)
Unconfirmed
- Whether Anthropic or Claude actually solved the Navier–Stokes problem remains unsupported by any official statement or paper
- Where the core ideas came from, how much guidance Claude needed, and whether the approach generalizes to other problems all remain unknown (questions raised in m6)
Why it matters
- Commenters like kimmonismus argue that if the rumor is true and the proof survives expert scrutiny, it would be one of AI's biggest moments ever, marking a watershed for AI-driven original research (m6)
- Tao's remarks triggered reflection on AI benchmark culture: he opposes math-competition-style research done "just for scores," arguing it could backfire on mathematics itself (m8, m9)
- The episode reflects the AI community's eagerness for proof of model capability; some users have called on Anthropic to publish such results as the best possible marketing (m1)
- Terence Tao warns AI-cracked math problems could pollute and stall mathematical progress — 量子位 · 2026-09-05
- Rumor: Claude may have solved Navier–Stokes, with paper out for expert review — inductionheads · 2026-09-05
- Gossip: Anthropic May Have Solved Navier-Stokes With AI, Unconfirmed — iskander · 2026-09-05
- Rumor: Anthropic's model solved a Millennium Prize Problem, Terence Tao reacts — IgorCarron · 2026-09-05
- Terence Tao responds to rumors that an AI lab cracked the Navier-Stokes problem — elsleightholm · 2026-09-05
- Report: Claude may have cracked the Navier–Stokes Millennium Problem, pending expert scrutiny — MoonL88537 · 2026-09-06
- Solver of Navier-Stokes would be Anthropic's best marketing, quips AI observer — burny_tech · 2026-09-06
- Rumor: Anthropic Solved a Millennium Prize Problem, Terence Tao Responds — littmath · 2026-09-06
- Terence Tao responds to Navier-Stokes AI rumor with warning against benchmark-driven math races — 量子位 · 2026-09-06
Episode 3 · Terry Tao slams closed labs' theorem-proving as viral marketing (2026-09-06, 2 posts)
Terry Tao, relayed by Delip Rao, criticized frontier labs' AI theorem-proving as viral marketing rather than real progress for mathematics, noting many mathematicians are denied access to those same models.
- Terry Tao slams frontier labs' theorem proofs as viral marketing, not math progress — deliprao · 2026-09-06
- Mathematicians say frontier labs stonewall their access to advanced AI models — deliprao · 2026-09-06
Episode 4 · Claude Did Not Solve Navier-Stokes, Clarifies Terence Tao (2026-09-06, 2 posts)
Rumors that Claude solved the Navier-Stokes equation are unfounded, with no papers or official evidence from Anthropic. Terence Tao clarified the claim stemmed from a hypothetical example he posted, and the Clay Millennium Problem remains unsolved.