For 90 years, one of mathematics' hardest questions sat unsolved, quietly resisting every attempt to crack it. Then, within the span of a single week in September 2026, an AI system may have solved it — and the story took a turn nobody expected: accusations that the win wasn't entirely OpenAI's to claim.
Here's what actually happened, why it matters, and why the mathematics community isn't celebrating quite yet.
📋 Table of Contents
What Is the Navier-Stokes Existence and Smoothness Problem?
The Navier-Stokes equation is one of the most important formulas in physics — it describes how fluids move, and it's used every day to design aircraft, model weather systems, and simulate blood flow through the human body. In practice, the equations work fine. Engineers use them constantly.
But there's a purely mathematical question buried inside them that no one has ever been able to answer: if you start with a perfectly smooth, well-behaved 3D fluid flow, does it always stay smooth forever — or can it "blow up," producing an impossible mathematical singularity?
This is known as the Navier-Stokes existence and smoothness problem, and it's one of the seven Clay Mathematics Institute Millennium Prize Problems — a list of the hardest unsolved questions in math, each carrying a $1 million prize since the year 2000. Only one of the seven has ever been solved.
OpenAI Says Its AI Solved It
On September 8, 2026, OpenAI announced that an internal, unreleased AI system had produced a proof showing that the Navier-Stokes equations can, in fact, develop a singularity in finite time — a major theoretical result. The company published both a full write-up of the proof and a formal verification of it using the Lean proof assistant, a tool mathematicians use to check that a proof is logically airtight line by line.
If verified and accepted by the mathematics community, this would be one of the most significant results ever produced by an AI system — arguably more significant than AI beating humans at chess or Go, because this is original mathematical research, not a game with fixed rules.
Then Came the Accusations
Almost immediately, the story got complicated. Tristan Buckmaster, a mathematics professor at NYU, and Levent Alpöge, a mathematician working at Anthropic, said they had been working on the exact same problem for nearly a year — and that they had reached their own breakthrough on August 15, 2026, weeks before OpenAI's announcement.
According to accounts of the dispute, Buckmaster and Alpöge had used AI tools extensively in their own research process, and their progress may have leaked or been picked up on before they could publish. Buckmaster reportedly published a hastily prepared version of their own results once OpenAI's announcement became public, along with a document describing what he characterized as questionable conduct.
OpenAI, for its part, has described its own process in detail: after hearing rumors related to the Buckmaster-Alpöge research in early September, a swarm of AI agents was set loose on the problem, reportedly reaching a resolution within about 88 hours, followed by roughly 17 additional hours of formal verification — a staggering amount of computation condensed into just a few days.
Why This Story Matters Beyond the Drama
Strip away the controversy, and there's a bigger question here: what happens to mathematical research when AI systems can independently produce PhD-level, publishable-quality proofs in days instead of years?
For decades, the Navier-Stokes problem has been considered one of the "holy grails" of mathematical physics — a problem so hard that entire careers have been built around partial progress on it. If AI systems can meaningfully contribute to problems at this level, it raises real questions about credit, collaboration, and what mathematical discovery will look like going forward.
Frequently Asked Questions
Is the Navier-Stokes Millennium Prize problem officially solved?
As of this writing, the Clay Mathematics Institute has not verified or officially accepted any proof of the Navier-Stokes existence and smoothness problem. OpenAI's result is undergoing scrutiny from the mathematics community, and formal acceptance — including the $1 million prize — typically takes significant time to confirm.
What does "singularity in finite time" mean?
It means that under certain conditions, a perfectly smooth fluid flow can theoretically develop a point where its behavior becomes mathematically undefined — essentially "breaking down" — within a limited period of time, rather than staying smooth forever.
Who are Tristan Buckmaster and Levent Alpöge?
Buckmaster is a mathematics professor at NYU who has worked on fluid dynamics problems for years. Alpöge is a mathematician who works at Anthropic. Both claim to have made independent progress on the same problem shortly before OpenAI's announcement.
Does this mean AI can now do original math research?
This case suggests AI systems are becoming capable of contributing to frontier-level mathematical research, though experts caution that formal verification, peer review, and independent confirmation are still essential steps before any result — human or AI-generated — is fully accepted.
Final Thoughts
Whatever the mathematics community ultimately decides about who deserves credit, one thing is already clear: the line between "AI as a tool" and "AI as an independent researcher" just got a lot blurrier. Expect this story — and the questions it raises — to keep evolving in the weeks ahead.
