
An Anthropic mathematician says he has proved the Riemann hypothesis with help from the company’s Claude AI system, a claim that would represent a major advance in mathematics if it withstands independent scrutiny.
Levent Alpöge has posted a proof written in Lean, a computer language used to formalize mathematical arguments. The work has not been independently verified, and Anthropic has not endorsed or publicly commented on his claim.
Now loading...
Alpöge wrote that he contacted Anthropic CEO Dario Amodei about the result late Sunday night. He also indicated that he waited to publish the work until after Anthropic’s initial public offering on Thursday.
First proposed in 1859, the Riemann hypothesis concerns the distribution of prime numbers and the zeros of a mathematical function. It’s among seven exceptionally difficult problems designated for the Millennium Prize, with a $1 million award for each solution. The Clay Mathematics Institute requires a proposed proof to undergo at least two years of scrutiny and gain broad acceptance among mathematicians before it can qualify for the prize.
Alpöge said Claude helped develop the argument and that the approach could also prove the generalized Riemann hypothesis. He said he and the AI had formalized the reasoning without relying on assumptions.
Anthropic said in August that it was using Claude to work on the problem, while cautioning that the methods involved were not expected to prove the hypothesis.
Only one of the seven Millennium Prize problems has been officially solved since the awards were established: the Poincaré conjecture. OpenAI said in September that it had solved another, the Navier-Stokes problem, which concerns fluid flow. That proposed solution has been viewed by some as credible but has not received formal recognition.
OpenAI said it devoted millions of dollars and 10,000 AI agents to its effort. The work has also prompted controversy over allegations that the company used mathematicians’ unpublished research shared with its AI tools. OpenAI has denied wrongdoing.
Alpöge previously said he solved the Jacobian conjecture with Claude. As AI systems increasingly contribute to research, mathematicians are working to assess the claims and understand how the technology may change the field.
