Mathematician Claims AI-Generated Proof of Riemann Hypothesis

Levent Alpöge, a mathematician at Anthropic, has announced a proof of the Riemann hypothesis using artificial intelligence, though experts warn verification will take years.
Levent Alpöge, a mathematician at artificial intelligence firm Anthropic, has announced that he has developed a proof of the Riemann hypothesis. The work, which reportedly utilizes the company’s AI chatbot, Claude, consists of 17 million lines of code—a volume of data that has not been fully reviewed by any human.
The Riemann hypothesis, which addresses the distribution of prime numbers, has been one of the most challenging problems in mathematics since it was first conjectured in 1859. As one of the Clay Mathematics Institute’s Millennium Prize Problems, any verified proof would represent a monumental advancement in the field.
However, the nature of the claim—relying on a vast, machine-generated output—has drawn caution from the scientific community regarding the verification process. Tony Padilla, a theoretical physicist at the University of Nottingham, noted that the proof’s complexity necessitates a rigorous human-readable translation.
“This can take years,” Padilla said regarding the validation of the mathematical output. “There may be problems and we don’t know. It’s very early days.”
Beyond the technical challenge of verification, Padilla also emphasized the necessity for transparency, particularly given the involvement of a private firm. “It’s absolutely crucial that everything is shared in a way that the community can follow,” he said, acknowledging that Alpöge and Anthropic would likely have “commercial interests at heart.”
The announcement follows previous interest in the potential of AI for mathematical research. Brian Conrey, an expert on the Riemann hypothesis who previously reviewed an earlier paper involving Claude for Anthropic in August, described that earlier moment as “the biggest thing that’s ever happened in mathematics.”
The current status of the proof remains preliminary, with experts waiting for a version that can be scrutinized by the global mathematical community.