Open LoopNEWS
ReportedModels · 2 min read

NYU and Anthropic researchers claim an AI-aided Navier-Stokes proof, with no outside force

On Friday 20 November 2026, NYU professor Tristan Buckmaster and Anthropic's Levent Alpöge posted a claimed AI-aided proof that the Navier-Stokes equations, which model fluids, can blow up in finite time with no outside force. It aims to close a loophole in OpenAI's September result. No outside expert has confirmed it yet.

The Anthropic logo
Image: Anthropic

The short version

  • 01Buckmaster (NYU) and Alpöge (Anthropic) posted the claimed proof on Friday 20 November 2026.
  • 02OpenAI's September proof used a finely tuned outside force. The new claim removes it.
  • 03The authors used Claude and Codex, and call the write-up unpolished and heavily LLM-assisted.
  • 04No independent expert has confirmed the proof.

At a glance

StatusReported (Scientific American, The Next Web). Not confirmed by outside experts
PostedFriday 20 November 2026
AuthorsTristan Buckmaster (NYU), Levent Alpöge (Anthropic)
ClaimNavier-Stokes can blow up in finite time with no outside force
Tools usedClaude and Codex, per The Next Web
Checked by outsidersNot yet

What is the Navier-Stokes claim?

The Navier-Stokes equations describe how fluids such as water and air move. On Friday 20 November, Tristan Buckmaster of NYU and Levent Alpöge of Anthropic posted a claimed proof that they can blow up: in the maths, a fluid's speed becomes infinite in finite time, even with no outside force stirring it.

Scientific American calls it a new AI-aided proof. The Next Web says it was posted on GitHub as a 79-page paper with a computer-checkable Lean version.

The Anthropic logo
Alpöge works at Anthropic. Image: Anthropic

How is it different from OpenAI's September result?

Scientific American says OpenAI's September proof met the technical conditions of the Clay Mathematics Institute's $1 million prize, but relied on a finely tuned outside force that many experts saw as a loophole. The new claim removes that force.

The two sides also had a public dispute in September. Buckmaster suggested OpenAI had scooped his work, and OpenAI denied any foul play.

Has anyone checked it?

Not independently. Buckmaster called the write-up "unpolished and heavily LLM-assisted", and The Next Web says a data file the paper relies on had not yet been released. Terence Tao of UCLA wrote that the strategy is broadly clear but a properly written proof still needs work.

Alpöge also claimed a proof of the Riemann Hypothesis the same day. We covered it in our [Riemann story](/news/anthropic-mathematician-claims-claude-riemann-hypothesis/).

Anthropic's illustration for its Riemann zeta research
Anthropic's earlier Riemann work. Image: Anthropic

Why does it matter?

If it holds, AI helped settle a problem mathematicians care about, for the second time in a few weeks. It would also show AI tools from two rival labs working on the same question. Earlier, [OpenAI published 722 maths manuscripts](/news/openai-722-math-manuscripts/) from an unreleased model.

What does it mean for you?

Nothing changes in your apps this week. This is a maths result, not a product, and it is a claim, not a confirmed proof.

What you can take from it: both OpenAI and Anthropic now say their AI helps with research maths. Treat any such claim as early until outside experts have checked it.

Compare it with our story on [Anthropic's Riemann claim](/news/anthropic-mathematician-claims-claude-riemann-hypothesis/): both are unchecked, and both lean on a computer-checkable Lean file.

One thing to do now: wait for expert checks before you share it as fact.

What's next?

Mathematicians will check the paper. The authors say the missing data file will be released.

Questions people ask

What does blowup mean?

In the maths, a fluid's speed becomes infinite in finite time. Real fluids cannot do this.

Is the problem solved?

Not confirmed. No outside expert has confirmed the proof.

Which AI tools were used?

The Next Web says the authors used Claude and Codex.

Sources

  1. PrimaryScientific American: After months of controversy, AI finally solved the 'real' Navier-Stokes problem (20 Nov 2026)
  2. CoverageThe Next Web: Navier-Stokes can blow up with no outside force, Buckmaster and Alpöge say (21 Nov 2026)

Why "Reported"? We label a story "Reported" when only the media have it so far. Spotted an error? Email hello@openloopnews.com and we will correct it.