OpenAI announced on September 8, 2026, that an internal artificial intelligence system had produced a proof of finite-time singularity formation for the three-dimensional Navier–Stokes equations. The company published a mathematical write-up and a formalization in Lean, a proof-verification system. The announcement prompted debate about the mathematical significance of the construction, the role of an external force, and a separate dispute over credit and related research.
The Navier–Stokes equations describe how fluids move and are important to fields including aerodynamics, weather forecasting and fluid mechanics. One of the central mathematical questions is whether a smooth solution for a three-dimensional incompressible fluid can develop a singularity in finite time, meaning that the velocity becomes unbounded. The Clay Mathematics Institute designated the existence-and-smoothness problem as one of its seven Millennium Prize Problems in 2000, with a $1 million award for a qualifying solution.
## What OpenAI announced
In its [September 8 announcement](https://openai.com/index/navier-stokes-solution/), OpenAI said its internal system had constructed a solution in which a fluid starting at rest develops a singularity despite viscosity, while its energy remains finite. The construction uses a smooth external force.
OpenAI said the Navier–Stokes effort began on September 1 and reached its result on September 5, about 88 hours after the first agents were launched. The company reported that the system used approximately 10,000 concurrent agents during the effort and that Lean formalization and verification required additional work. OpenAI described the result as establishing statements C and D in the official Millennium Prize formulation. It also stated that it did not intend to claim the $1 million prize for this result.
These are OpenAI’s claims about its own work. The release of a proof and its formalization provides material for independent examination, but readers should distinguish the company’s announcement from a final determination by the Clay Mathematics Institute.
## Why the external force matters
The debate requires a distinction between the different versions of the official problem. Charles Fefferman’s [Clay Mathematics Institute problem description](https://www.claymath.org/wp-content/uploads/2022/06/navierstokes.pdf) sets out four alternatives. Statements A and B concern existence and smoothness with the external force set to zero; statements C and D concern breakdown and permit a smooth external force subject to specified conditions.
Consequently, the mere presence of an external force does not prove that a construction falls outside the written problem statement. The substantive question is what the construction establishes, how its force is defined, and what that result means for the broader regularity problem that has attracted mathematicians for decades.
Some experts have questioned whether a construction relying on a specially designed force answers the version of the problem most people have in mind when discussing the possible breakdown of fluid motion. [Scientific American reported on September 21](https://www.scientificamerican.com/article/did-openai-solve-the-wrong-navier-stokes-problem/) that the external-force issue had become a central part of this debate.
A [September 17 paper by Peter Constantin, Mihaela Ignatova and Vlad Vicol](https://arxiv.org/html/2609.20803v1) provides additional technical context. The authors study regularity for solutions under assumptions that include real-analytic forcing and particular properties of the velocity field. They show regularity under those assumptions and explain how the properties of OpenAI’s construction relate to their result.
The paper describes the force in OpenAI’s construction as smooth but not real-analytic. Its theorem therefore does not, by itself, disprove OpenAI’s construction: the relevant assumptions differ. Instead, the paper illustrates why the precise properties of the forcing and solution must be examined when assessing what the construction establishes.
## The dispute over credit and related research
The announcement also became entangled in a dispute involving Tristan Buckmaster, a mathematician at New York University, and Levent Alpöge, an Anthropic researcher who was working independently on the mathematics. Their related work concerned finite-time singularity formation for Euler equations, which are closely related to the Navier–Stokes equations but omit the viscosity term.
Buckmaster publicly described discussions with OpenAI about the timing of the work, how related contributions would be credited, and whether information from their Codex sessions could have influenced OpenAI’s system. His account raised concerns about how the proposed credit arrangements would treat Alpöge’s contribution. These statements describe Buckmaster’s account of the discussions, not an independently established finding of misconduct.
OpenAI provided a different account in its [announcement and concurrent-work update](https://openai.com/index/navier-stokes-solution/). The company said its system did not access the researchers’ unpublished work before it became public and that its investigation found their prompts could not have influenced the result. OpenAI acknowledged that it could not entirely rule out an indirect influence from de-identified usage data used to improve its models. It also recognized the priority of Buckmaster and Alpöge’s work on the related forced Euler problem.
The researchers’ related results, OpenAI’s Navier–Stokes construction and the dispute over interactions between the people involved are distinct questions. Evidence about one does not, on its own, settle the others.
## What the official status means
On September 11, the [Clay Mathematics Institute responded to the announcement](https://www.claymath.org/news/navier-stokes-announcement/), saying the problem had apparently been settled while emphasizing that the evaluation and prize process would be deliberately unhurried as the work was analysed and scrutinized.
As of October 28, 2026, the institute’s [Navier–Stokes problem page](https://www.claymath.org/millennium/navier-stokes-equation/) continues to display the status “Active.” The public status therefore does not establish that the $1 million prize has been awarded for OpenAI’s result.
The accurate description is that OpenAI has announced a proof and made material available for examination, while the mathematical interpretation and significance of the construction remain subjects of scrutiny. The external-force issue should be explained in the context of the actual Clay formulation, and the credit dispute should remain attributed to the people and organizations involved.
The development is significant for AI-assisted mathematical research, but the announcement, independent mathematical assessment and official prize determination should not be treated as interchangeable.
