Two rival claims to a proof of the Riemann hypothesis, the most famous unsolved problem in mathematics, arrived within hours of each other on Friday, and neither has been verified.
Levent Alpöge, a mathematician at Anthropic, said he and Claude had found a proof, and that the same method settles the generalised version. In a four-page note, he wrote: “I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis.”
The proof was released only as a Lean formalisation, in which computers check mathematical arguments step by step. There is no conventional paper yet: a full written argument would run to more than 250,000 pages.
Independent mathematicians report the Lean code compiles without errors, including Andrew Sutherland of MIT. But compiling is not understanding: no human has read the proof in full, and the Clay Mathematics Institute lists the problem as unsolved.
A second proof, written by a human alone
Hours later, Ravi Vakil of the American Mathematical Society published a rival draft by Stanford postdoc Alessandro Masullo. His 940-page proof, 12 years in the making, was written by a human. “We are still here. We are very much alive. We are not going anywhere,” he wrote.
Masullo has acknowledged mistakes, other mathematicians have reported problems, and Brian Conrad, who checked the draft for months, said more work was needed. “To judge the success of the method requires the paper to reach a stage of more final revision,” he said.
Why the claim matters, and why caution is needed
Proposed by Bernhard Riemann in 1859, the hypothesis describes hidden order in the distribution of prime numbers. It is one of seven Millennium Prize Problems, each worth a million dollars, and a solution must appear in a peer-reviewed journal and withstand two years of scrutiny.
It came the day after Anthropic’s stock market debut, after its filing to list its shares. Alpöge had consulted chief executive Dario Amodei first, and acknowledged the timing could look cynical. The claim rests on Claude, whose latest release, Claude Fable 5.2, promises faster answers.
Two months ago, a rival lab said one of its models had solved the Navier-Stokes equations, another Millennium Prize problem, and mathematicians say that proof appears correct but is not fully vetted.
Brian Conrey, a specialist, called it “a monumental moment in human history”, though he had not seen a manuscript. James Maynard said: “I suspect that the proof will turn out to be correct, although even checking the Lean verification will take some time.”
This may change what a proof looks like: software can check an argument no one can read in full. But the prize rules demand a readable paper, peer review and two years of scrutiny, so the prize is nowhere near claimed.


