An Anthropic mathematician says he used the company's Claude chatbot to produce a complete proof of the Riemann hypothesis, a nearly 170-year-old problem widely considered the most famous unsolved question in mathematics — a claim that, if verified, would rank among the most significant results in the field's history, but which outside experts caution is nowhere near confirmed.
Levent Alpöge said Friday, a day after Anthropic completed its first week of trading on the Nasdaq following its IPO, that over roughly six weeks he worked with a non-public version of Claude to construct a detailed proof of the hypothesis. The full written proof, produced in Lean, a computer language mathematicians use to formally verify proofs, runs more than 250,000 pages.
A problem at the heart of mathematics
Named for 19th-century mathematician Bernhard Riemann, the hypothesis concerns the distribution of prime numbers and has eluded top mathematicians for generations despite its outsized importance: because prime numbers underpin much of higher mathematics, a genuine proof would ripple across the field. The claim follows a string of announcements in recent months from AI labs, including OpenAI, touting progress on major unsolved math problems.
Alpöge said he messaged Anthropic chief executive Dario Amodei directly on Sunday to ask how best to announce the finding, acknowledging the timing — arriving just a day after Anthropic's market debut — looked suspicious. "The precise timing is horrible," he said. "If I could have, I'd have finished this all much earlier. I am clearly unilaterally responsible for feeling I just could not live if I waited even longer. There is an obvious cynical view, to do with the timing, but it is false." He described the potential result as "one dawn across science of something far greater."
Verification could take months
Terence Tao, one of the most prominent mathematicians working today, said Friday that verifying the proof is still some way off — a process that, given the proof's enormous length and the difficulty of the underlying problem, is likely to involve painstaking review by multiple independent experts before the mathematical community reaches any consensus. Anthropic has not claimed the proof is confirmed, and the company's own August announcement of a related, more modest result — improving a bound on the fraction of Riemann zeta zeros lying on the critical line — was explicitly described by outside mathematicians as not a path toward resolving the full hypothesis.
Given the scale of the claim and the history of false alarms around the Riemann hypothesis, mathematicians said the announcement should be treated as a serious but entirely unproven claim until independent verification is complete, which could take months or longer.