Featured Post

Mathematicians Claim AI Helped Prove the Real Navier Stokes Problem With No Outside Force

Swirling cyan fluid vortex with a fine grid on a white background

Mathematicians Tristan Buckmaster and Levent Alpoege posted a proof on 20 November 2026 claiming that the Navier Stokes equations can blow up in finite time with no outside force at all. This is the version of the problem many mathematicians call the real one, because OpenAI's September proof needed a smooth outside force to push the fluid, and this claim says no force is needed. The claim is not verified yet, and the key piece of evidence behind it has not been released.

The paper is 79 pages long and comes with a Lean 4 formalization of about 600,000 lines across roughly 2,270 files. The authors say they used Anthropic models, Claude, throughout the work, and Buckmaster described the write up as unpolished and heavily LLM assisted, written in a rush after the September dispute. Until outside mathematicians check the work, this is a claim, not a confirmed proof.

What happened

On 20 November 2026, Buckmaster of New York University and Alpoege of Anthropic posted a paper titled Finite time blow up for the Navier Stokes equations. It claims that smooth starting flows exist whose solutions to the three dimensional Navier Stokes equations blow up in finite time with no force, which would disprove the Clay Mathematics Institute statements that smooth solutions always last. The same day brought a second paper with Matei Coiculescu showing the same kind of blowup for the Euler equations.

The paper leans on a numerical certificate, a set of computed numbers that proves a key profile exists. That certificate was not released with the paper. The paper gives only a cryptographic hash of it and says it will be released soon. Without the certificate, no outside mathematician can check the central computation yet.

The Lean formalization has its own gaps. The repository notes say a patch changed some statements on 20 November and the patched package was not rebuilt when the repository was assembled. The proof library itself contains no sorry placeholders, the markers Lean uses for unproven steps.

ClaimDateForce neededCertificate status
OpenAI8 September 2026Smooth outside forceReleased with the paper
Buckmaster and Alpoege20 November 2026NoneNot released, hash only
Anthropic official logo

Official Anthropic logo. Credit: Anthropic.

Why it matters

If the claim holds, it closes the loophole in the September result. OpenAI proved blowup with a force pushing the fluid, and critics noted that a real fluid in the prize problem needs no such push, so the unforced case was the one that mattered. Buckmaster and Alpoege say their proof covers both open space and a periodic box, for every positive viscosity. See also our post on the Claude Riemann Hypothesis proof claim from the same day.

The claim also shows how fast AI assisted mathematics now moves. The same two mathematicians were at the center of the September priority dispute with OpenAI, and two months later they posted a bigger claim built with Claude throughout, according to the paper itself. The dispute over who deserves credit is now a race over who can verify first.

But the missing certificate is the whole story for now. No independent mathematician quoted in the coverage has said the proof is correct, and the key computation cannot be checked until the certificate appears. Clay rules also require publication and two years of general acceptance before any prize.

What to watch

Watch for the certificate. The authors promised it will be released soon, and the moment it appears, independent mathematicians can rerun the interval arithmetic check that the whole proof depends on. Until then, treat every headline about this as a claim.

Watch Terence Tao's blog and the Lean community too. Tao has already listed the claim alongside the Riemann claim from the same day, and the Lean formalization will face the same kind of line by line machine check. The first real signal will be whether the patched package builds cleanly.

Follow the verification

No official video covers this claim yet, and the paper itself is the primary source. The verification status is tracked on the Navier Stokes existence and smoothness page.

Sources

Wikipedia: Navier Stokes existence and smoothness, including the 20 November 2026 claim

Nacionale: Unforced Navier Stokes blowup claimed, what it means

AI Weekly: Alpoege, Buckmaster post AI aided unforced Navier Stokes blowup

OpenAI: Finite time blow up for Navier Stokes, the September forced case paper

Follow Learn Daily AI on Instagram at @learningdailyai for one new AI story every day.

Comments