Scientific American reported on Friday that Levent Alpöge, a mathematician who works at Anthropic, claims he proved the Riemann hypothesis using Claude. Alpöge wrote on his website that he believes he and Claude found a proof. Bernhard Riemann first proposed the problem in 1859. It says the positions of prime numbers follow a hidden order. Alpöge posted the proof in Lean, a programming language that checks each step of a mathematical argument. He says it also confirms the Generalized Riemann Hypothesis, a much wider version of the problem.
The proof has not been confirmed. Scientific American says problems can arise in Lean itself. Experts must check that the problem was coded correctly and that the proof holds. The outlet says this should not take long, but it gives no date. It describes the story as breaking news that will be updated. The article does not say how Claude was used, or how much of the work was Claude's and how much was Alpöge's.
For people who build or buy software, this is a claim to watch, not a result to rely on yet. Scientific American says that if the proof is correct, it would show how strong AI reasoning has become. Expert review of both the Lean version of the problem and the proof itself will decide whether the claim stands.