Levent Alpöge, a mathematician at Anthropic, claims he has found a proof of the Riemann Hypothesis working with the company’s Claude AI model, according to a note posted Friday on his personal website. “I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis,” the note says. Scientific American reported the claim Friday.
The Riemann Hypothesis, proposed by Bernhard Riemann in 1859, is a conjecture about where the zeros of the Riemann zeta function lie, which governs how prime numbers are distributed. Scientific American calls it the most famous open problem in math.
The note says the method also resolves the Generalized Riemann Hypothesis, and that the author expects, “with more time, the Extended Riemann Hypothesis too.” It says, “Claude and I have also formalized the arguments without assumptions,” in Lean, a programming language in which a computer checks each step of a proof. There is no paper. The note calls the argument “incredibly long at the moment” and says the author decided to present only the Lean code, with a proper write-up to come.
The Lean formalization is posted on the same site as 14 compressed archive parts, about 240 megabytes in total. In the sources reviewed, no one had reported independently compiling the archive or auditing whether its final theorem states the Riemann Hypothesis. Scientific American wrote that it “shouldn’t take long for experts to verify whether the hypothesis is properly coded, and if the proof is correct.”
The four-page note does not carry an author’s name; Scientific American attributed it to Alpöge. It says Alpöge guided the work toward the right setting and that “Claude far outpaced me near the finish line,” crediting Claude with steps including the length of a chain of “theta lifts.” The author writes that he has “little mastery of automorphic forms on higher rank groups” and that descriptions of parts where “Claude is the expert” are paraphrased from an AI summary “which might make zero sense.” A timeline the note says Claude compiled runs from Oct. 6 to Nov. 16.
In August, Anthropic said an unreleased version of Claude had raised a lower bound on the share of zeta zeros known to satisfy the hypothesis to 67.2% from 41.6%, work Alpöge examined. “We don’t expect that the techniques Claude used will lead to proving the Riemann hypothesis,” the company wrote then. Anthropic has also said Claude formalized Fermat’s Last Theorem in Lean.
Anthropic had not commented on the claim in the sources reviewed, including its newsroom and its X account, as of Friday morning.