NEWS20 November 2026Scholars say that the claim to have derived a proof for a landmark problem in mathematics will need rigorous checking.ByDavide CastelvecchiDavide CastelvecchiView author publicationsSearch author on: PubMed Google ScholarView saved researchThe Riemann hypothesis has been a lodestar for generations of mathetmaticians. Credit: James Brittain/GettyIn a year that has seen artificial intelligence reach one spectacular mathematical breakthrough after another, Anthropic is now claiming the top prize of them all: a full proof of the Riemann hypothesis, the statement that many mathematicians consider the most important open problem in the field.The announcement came from Levent Alpöge, a mathematician at the San Francisco, California company. “I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis,” Alpöge wrote, referring to Anthropic’s large language model (LLM).“If this fully checks out, it is clearly a historic milestone in mathematics,” says number theorist James Maynard, a Fields Medal winner at the University of Oxford, UK, says.It would also be a spectacular bicentenary present for Bernhard Riemann, the German mathematician who was born 200 years ago this September and who formulated the problem, which addresses the distribution of prime numbers, in the mid-1800.Prime problemThe problem has to do with the ‘Riemann ζ function’ (ζ is the Greek letter zeta), which encodes the randomness properties of the prime numbers. Riemann discovered that the distribution of primes is encoded in the ζ function’s ‘zeroes.’ These are points on a plane denoted by z satisfying the equation ζ(z) = 0.Riemann conjectured that apart from some obvious but uninteresting cases, all such solutions had to lie on a certain vertical line in the plane. If this were true, he showed, it would imply that prime numbers are scattered among all whole numbers in a way that is essentially as random as it could be.Alpöge’s 4-page document includes a sketch of the proof. But Andrew Granville, a number theorist at the University of Montreal in Canada, says that Alpöge should have taken the time to prepare a clearer write-up before making the announcement, which came just days after Anthropic’s initial public offering (IPO). The IPO initially valued the company at US$1.5 trillion, but the company’s valuation dropped on 19 November.“What he released is garbled,” Granville says of Alpöge’s document. He cites the downturn in the company’s stock value as the reason for the timing of the proof’s release.“There is an obvious cynical view, to do with the timing, but it is false,” Alpöge wrote in the document.Anthropic did not respond immediately to Nature’s request for comment.A dawn of science?Tristan Buckmaster, a mathematician at New York University who collaborated with Alpöge on a separate project regarding the Navier-Stokes equation, says that Alpöge deserves a lot of credit for developing techniques to produce mathematical results using LLMs. “Levent himself built this amazing setup to do math”. He adds that Alpöge, who is himself a number theorist on leave from Harvard University in Cambridge, Massachusetts, has been working relentlessly for months on this and other problems, hardly ever getting any sleep.‘It is incredible’: How AI is transforming mathematics“I suspect that the proof will turn out to be correct,” Maynard says, even though it is now in the form of incomprehensible code in the Lean programming language.“At the moment, it apparently is crazily complicated. It’s going to take a long time for mathematicians to digest,” says Buckmaster. Maynard agrees, but he adds, “assuming we do get the corresponding understanding, this will surely be viewed as one of the absolute pinnacles of mathematical results”.Alpöge wrote that in the proof, Claude conjured “an astounding amount of mathematics”. He wrote that he has dreamt of solving this problem since he was a student and waxed poetic about the significance of this moment. “It is natural to feel that this is an end, in my case a loss of a principal guiding light, but what I saw during the course of progress so amazed me that I believe it is one dawn across science of something far greater.”doi: https://doi.org/10.1038/d41586-026-03812-7SubjectsMathematics and computingMachine learningLatest on:Mathematics and computingMachine learningJobs Associate or Senior Editor (Computational and mathematical biology), Nature CommunicationsJob Title: Associate or Senior Editor, Nature Communications (Computational and mathematical biology) Location: Shanghai, Nanjing, Beijing, Pune,...Shanghai, Nanjing, Beijing, Pune, New Delhi, Hybrid working modelSpringer Nature LtdAssociate or Senior Editor (Biomaterials), Nature CommunicationsJob Title: Associate or Senior Editor, Nature Communications (Biomaterials) Location: Shanghai, Nanjing, Beijing, Pune, New Delhi, Hybrid working...Shanghai, Nanjing, Beijing, Pune, New Delhi, Hybrid working modelSpringer Nature LtdEmerging Scholars Forum 2027 | 19-20 January 2027 | SingaporeThe College of Engineering at NTU invites outstanding senior graduate students and postdoctoral fellows to apply for the Emerging Scholars Forum 2027.Singapore (SG)Nanyang Technological University, SingaporeTT Faculty - Protein Engineeringdevelops and applies innovative computational tools, particularly AI/ML-driven approaches to protein engineering and small-molecule discoveryCleveland, OhioCase Western Reserve University/PharmacologyAssistant/Associate Professor - PharmacologyThis is a renewable, 12-month, non-tenure-track position, with the majority of time devoted to teaching and advising undergraduate students.Cleveland, OhioCase Western Reserve University/Pharmacology