Press Lantern

News, held to the light

Science

An Anthropic Mathematician Says Claude Helped Him Prove the Riemann Hypothesis. Mathematicians Are Skeptical.

Levent Alpöge's claim to a 167-year-old Millennium Prize problem — made days after his employer's record IPO — is one of two extraordinary mathematics assertions he posted this week. Neither has been independently verified.

Editorial illustration of a chalkboard filled with mathematical notation dissolving into points of light above a quiet desk lamp.
Two Millennium Prize claims in one week — one attributed partly to a chatbot — remain unverified by independent mathematicians. AI-generated illustration for Press Lantern

An Anthropic researcher has claimed he used the company's Claude chatbot to prove the Riemann hypothesis, the 167-year-old conjecture about the distribution of prime numbers that carries a $1 million Millennium Prize — an assertion circulating widely on Friday that no independent mathematician has yet confirmed.

Levent Alpöge said on November 20 that Claude had helped him prove the hypothesis, which has resisted every serious attempt since Bernhard Riemann posed it in 1859, according to The World of AI, which tracked coverage across thirteen outlets. The claim drew immediate scrutiny rather than celebration: a verified proof of a Millennium Prize problem would be one of the mathematical events of the century, and extraordinary claims are ordinarily weighed for months or years before acceptance.

It is not even the week's most specific claim to carry Mr. Alpöge's name. He and Tristan Buckmaster posted a 79-page paper and a Lean formal-verification package to GitHub asserting finite-time blowup for the unforced Navier-Stokes equations — effectively claiming to solve a second Millennium Prize problem, explainx.ai reported. Scientific American covered the Navier-Stokes claim, but the repository's own README notes the numerical certificate is not included and that the patched Lean package had not been rebuilt when the repository was assembled; no independent mathematician quoted by the magazine has confirmed the result.

Posted, unverified, and — if either survives — historic.

The twin claims arrive at a conspicuous moment. Anthropic completed the largest initial public offering in history on Thursday, and a Claude-assisted proof of the Riemann hypothesis would be the single most powerful advertisement imaginable for the company's models — a coincidence of timing that skeptics noted across social media and that the claimants have not addressed.

Specialists urged patience over either dismissal or acceptance. Machine-checked proofs in systems such as Lean can, in principle, settle questions like these decisively, but only once the certificate is complete and the kernel is run by others. Until then, both claims sit in limbo: posted, unverified, and — if either survives — historic.

The episode also marks a milestone of a different sort. Whatever the outcome, it is among the first times a major unsolved problem has been publicly claimed with a commercial AI system named as a collaborator, raising questions about credit, verification and the Clay Mathematics Institute's rules, which require publication and two years of general acceptance before any prize is awarded.

mathematics Riemann hypothesis AI Anthropic

← Back to the front page