Anthropic Researcher Says Claude Helped Prove the Riemann Hypothesis, Unverified
Levent Alpoge released a four-page note and a Lean proof on Friday; outside mathematicians have not yet checked it, and the claim comes days after Anthropic's record IPO.

A mathematician at Anthropic said Friday that he used the company's Claude chatbot to prove the Riemann hypothesis, a problem that has resisted mathematicians for more than 160 years. The claim has not been independently verified, and outside experts have not yet reviewed the proof.
"I believe I have, working with Claude, discovered a proof of the Riemann hypothesis," Levent Alpoge wrote in a four-page note released Friday. He said the method also settles a broader version known as the generalized Riemann hypothesis.
The hypothesis was proposed by the German mathematician Bernhard Riemann in 1859. It concerns how prime numbers, the building blocks of arithmetic, are distributed along the number line, and many mathematicians regard it as the most important open question in pure mathematics. It is one of seven Millennium Prize Problems chosen by the Clay Mathematics Institute in 2000, each carrying a $1 million prize. Under the institute's rules, a solution must be published in a peer-reviewed journal and withstand two years of scrutiny before it can win.
Alpoge said he and Claude produced a version of the proof in Lean, a programming language that lets a computer check every step of a mathematical argument. He is releasing only that version for now. A full written argument would be extremely long, he said, and he has not yet mastered all of the mathematics involved. He plans to write it up properly later.
That detail matters. A proof verified in Lean does not rely on a human reader catching errors, but the check is only as good as the statement that was formalized, so experts will want to confirm that the Lean statement matches the real Riemann hypothesis. Alpoge acknowledged the question of credibility in other ways. He said he consulted Anthropic chief executive Dario Amodei before announcing the result, and he conceded the timing could be read cynically without elaborating. He was probably referring to the company's record initial public offering this week. Anthropic did not immediately respond to a request for comment from AFP.
The company has made smaller claims before. In August, Anthropic said an unreleased version of Claude had improved a long-standing result related to the Riemann hypothesis, while stressing that it had not proved it.
Friday's claim also follows a similar announcement by a rival. Two months ago, OpenAI said one of its AI models had solved another Millennium Prize Problem, concerning the equations that govern how fluids move. Mathematicians say that proof appears correct but has yet to be fully vetted. Alpoge, who had worked on related fluid problems with New York University mathematician Tristan Buckmaster, was at the center of a dispute over credit for that work. OpenAI denied wrongdoing.
For now, the proof is an announcement, not an established result. The real test comes when specialists in number theory read the Lean files and the eventual written argument, and when the Clay institute's multi-year process runs its course. If it holds, it would be the first Millennium Prize Problem in pure mathematics solved with a large role for an AI system.



