Levent Alpöge said on Friday that he and Claude discovered a proof of the Riemann hypothesis, the most famous open problem in mathematics. He posted the argument in Lean, a language that checks each step for mathematical correctness.
The claim is unusually direct: “I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis.” Alpöge is a mathematician and an employee at Anthropic, and his statement puts an AI model inside a result that would reshape number theory if it holds up.
Claude and Lean
Lean matters here because it does not just display a proof; it lets experts inspect whether the logic actually works. That gives the claim a path from announcement to verification, but it also means the burden now shifts from wording to code-level correctness.
Alpöge said his proof confirms the Generalized Riemann Hypothesis as well. He also wrote, “It is natural to feel that this is an end, in my case a loss of a principal guiding light,” and added, “I hope the argument provides a glimpse towards the miracles that lie ahead.”
Bernhard Riemann and number theory
Bernhard Riemann first proposed the Riemann hypothesis in 1859, and the conjecture has since become the North Star for number theory. Progress on it has been so rare that most top experts have spent little time working on it, which is part of why any serious claim draws immediate scrutiny.
Alpöge’s name is not new to this corner of mathematics. In July, he tweeted a Claude-generated solution to the Jacobian conjecture, a reminder that he has already been posting machine-assisted results in public.
What experts will check
The next step is simple in principle and unforgiving in practice: experts will have to verify that the Lean-coded proof is correct, line by line. If it is, Friday’s claim would move from a striking announcement to a result with consequences far beyond one proof.
That is where the story now sits. Alpöge says the argument exists, and the mathematical community will decide whether the code really closes a 167-year-old problem.







