An employee of the artificial intelligence company Anthropic says he used its Claude chatbot to prove the Riemann hypothesis, widely considered the most famous problem in mathematics. Levent Alpöge, who is both an Anthropic employee and a mathematician, made the claim on Friday, a day after the company made its US stock market debut.
Alpöge said that over about six weeks he used Claude to build a detailed proof. He said the proof was found using a non-public version of the chatbot. If it is confirmed as correct, the result would rank among the most significant in the history of mathematics, and Alpöge described it as “one dawn across science of something far greater”.
What the Riemann hypothesis concerns
The hypothesis is nearly 170 years old and deals with prime numbers. It is named for the mathematician Bernhard Riemann and proposes that prime numbers follow a determined order along the number line. Its difficulty is a large part of why it is seen as so famous.
Also read: Anthropic poised for $1.4tn Nasdaq stock market debut · Asian stocks fall as Anthropic debut fails to ease AI fears
Prime numbers form the foundation of higher mathematics, so any breakthrough related to them can have wide effects. The claim comes as AI firms such as OpenAI have announced scores of solutions to major higher maths problems in recent months, problems that had eluded scholars for decades.

Timing of the announcement and verification of the proof
Alpöge said he messaged Anthropic chief executive Dario Amodei on Sunday to ask him directly how he could announce the discovery. He called the timing “horrible”, referring to the company’s public stock listing. “If I could have, I’d have finished this all much earlier,” he said.
Alpöge also said he could not wait any longer and acknowledged that some would see a cynical motive in the timing, which he said was false. The complete proof, written in Lean, a computer language used to confirm maths proofs, is more than 250,000 lines long.
Terrance Tao, a prominent mathematician, said on Friday that verifying the proof is still some way off. “I imagine the digestion process will take some time, though in this case at least we have a human author to communicate with to assist with this task,” Tao said.
Scrutiny of earlier AI claims in mathematics
No comparable human author was obviously involved in major AI-assisted maths proofs earlier this year. OpenAI said in September that employees using a non-public version of its AI tools solved part of a notoriously difficult problem in just 88 hours. That problem belongs to the Navier-Stokes equations, which concern how fluids move.
More than 10,000 AI agents, which are computer programmed bots that operate somewhat autonomously, were set to the task as an AI training exercise. The resulting proof has yet to be confirmed as accurate and drew criticism from academic and higher maths communities.
A group of maths scholars said AI companies boasting of major solutions to show how sophisticated their products had become was “detrimental to the science of mathematics”. In a group letter, they warned: “We are witnessing a general threat to intellectual work, with misalignment between the outcome of the use of AI and its initial purpose.”
Anthropic has also claimed progress and solutions in mathematics and physics. A month after its September claim, OpenAI released what it said were solutions, proofs or major progress on 372 more outstanding higher maths problems. Experts will need months, if not longer, to review those claims.