Levent Alpoge says Anthropic’s Claude produced a proof of the Riemann hypothesis
Anthropic mathematician posted a four‑page explanation and a Lean formalization of the claimed proof; verification by the mathematical community will take time.

Levent Alpoge, a mathematician who works at Anthropic, announced on 20 November that he and the company's AI assistant Claude have produced what he claims is a proof of the Riemann hypothesis, a central unresolved problem in mathematics. Alpoge posted a four‑page explanation on his website and says the formal proof is embedded in code written in the proof language Lean.
What Alpoge says he did and how the proof is stored
Alpoge says he arrived at the discovery late on 15 November and informed Anthropic CEO Dario Amodei that he wanted to publish it as the company prepared to make a high‑profile public debut on the stock market. According to Alpoge, the proof appears inside some 17 million lines of Lean code — a format used to write fully formalized mathematical proofs that can be checked by proof assistants.
The Riemann hypothesis, first proposed in 1859, concerns the distribution of prime numbers as they grow toward infinity. It is one of the most famous open problems in mathematics and was named by the Clay Mathematics Institute as one of seven Millennium Prize Problems in 2000, with a $1 million award for a correct solution.
Prominent mathematician Alex Kontorovich said that, typically, significant mathematical discoveries are presented as formal papers that allow other researchers to follow and verify each step. He argued that a conventional paper provides a way for experts to understand why a claim is true.
Kontorovich added that Claude had checked the Lean code and confirmed the correctness of the formal development. He also cautioned that computer verification does not always mean a human fully understands every step: "Usually, when someone gives you a scientific paper, it is, in some sense, a 'code' for the proof. You have to read the paper, follow the derivation yourself and decide whether you believe every single line in it. And when you finish, you should deeply understand why a particular claim is true."
He continued: "Does that mean I understand it? Not yet. But the computer is extremely picky — much more picky than I am — so if you manage to satisfy the computer, there is a very, very large chance you will satisfy me as well. Still, I have to finish the work."
For now, the Clay Mathematics Institute still lists the Riemann hypothesis as unresolved. Martin Bridson, the Institute's president, has said that the process of evaluating a claimed solution takes time, and formal verification and community scrutiny are typically required before a problem is declared solved.
Photo: AP)


