AI Models Help Prove Navier-Stokes Blows Up in Finite Time

On November 20, 2026, NYU mathematician Tristan Buckmaster and Anthropic researcher Levent Alpöge posted a 79-page paper claiming a finite-time blowup proof for the three-dimensional Navier-Stokes equations without external forcing. The work, officially released on November 20, 2026, includes a massive 600,000-line Lean 4 formalization and represents one of the most significant AI-assisted mathematical breakthroughs attempted to date. The result targets a problem worth $1 million from the Clay Mathematics Institute, though peer review and critical verification components remain pending.
The researchers openly acknowledge their heavy reliance on large language models from both Anthropic and OpenAI, including Codex, to generate both the paper text and the formal verification code. Buckmaster described the write-up as "unpolished and heavily LLM-assisted" in a Mastodon post hours after publication, explaining the rushed timeline was motivated by wanting to stake a claim after an incomplete September 2026 attempt by OpenAI researchers that relied on unphysical forcing mechanisms.
What the Proof Claims
The Navier-Stokes equations describe fluid flow and have resisted complete mathematical understanding for over a century. The specific claim in this paper is that solutions to the three-dimensional equations with no external force can blow up, meaning certain physical quantities become infinite in finite time. This formulation is widely considered the hardest and most important version of the problem. According to the paper, once a numerical profile is established, the proof follows established computer-assisted techniques that have been validated in related problems.
The Lean 4 formalization spans approximately 2,270 files built on the Mathlib library, representing months of machine-generated formal verification work. Lean is a proof assistant that checks mathematical arguments line by line, ensuring logical consistency. The scale of this formalization dwarfs most human-written formal proofs, though questions remain about whether LLM-generated code introduces subtle errors that escape automated checking.
Missing Numerical Certificate Raises Questions
The proof's credibility hinges on a numerical certificate that was not released with the paper. Instead, the authors provided only a cryptographic hash and a promise that the artifact will be made available soon. This certificate contains the specific numerical profile the analytical proof depends on, essentially the starting conditions that lead to blowup. Without it, independent mathematicians cannot verify the central computational claim.
Scientific American's coverage noted that no outside mathematician had publicly vouched for the result at the time of publication. The absence of the certificate is particularly significant because computer-assisted proofs require both the analytical argument and the numerical data to be scrutinized together. The hash provides a commitment mechanism but does not substitute for actual verification.
Release Date and Context
Officially released on November 20, 2026, the paper arrives in a competitive landscape where multiple research groups are racing to solve millennium-prize problems using AI assistance. Buckmaster's reference to "last time" points to September 2026, when OpenAI-affiliated researchers posted a Navier-Stokes result that the mathematical community rejected because it used artificial forcing terms that do not appear in the pure problem statement. The current work explicitly addresses the no-force case, raising the stakes considerably.
The timing also reflects how AI tools are compressing research cycles. What might have taken years of manual proof-writing and formalization was accomplished in months, though at the cost of polish and immediate verifiability. The authors' candor about their methods, including listing which AI models contributed to which sections, sets a transparency standard that may become necessary as LLM-assisted research grows more common.
What This Means
If verified, this proof would resolve one of the seven Clay Millennium Prize problems and demonstrate that AI can contribute materially to breakthrough-level pure mathematics. The heavy LLM involvement shows both the potential and the pitfalls of machine-assisted research: rapid production balanced against incomplete verification artifacts and community skepticism. The mathematical community now waits for the numerical certificate, peer review, and independent confirmation before the $1 million prize and historical recognition can be considered. Whether this result stands or falls, it marks a turning point in how computational tools interact with theoretical research at the highest level.
on Emergent today





