News
November 22, 2026 · The Verge

Anthropic mathematician posts a Claude-assisted Lean proof of the Riemann hypothesis

Levent Alpöge says he proved the Riemann hypothesis with Claude and released the proof in Lean. The claim still faces review, and a rival human draft drew early objections.

Levent Alpöge, a mathematician who works at Anthropic, said on November 20 that he has proved the Riemann hypothesis with the help of Claude. The Verge, Scientific American and The Decoder report that he posted the announcement and a proof written in the Lean proof language on his website. Within hours, Stanford professor Ravi Vakil shared a separate, all-human draft proof of a related result.

What's new

  • The claim: in his statement, Alpöge wrote: "I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis." He says the method also settles the Generalized Riemann Hypothesis.
  • Format: according to The Decoder, the proof was released only in Lean, split into 14 packages, and Alpöge says a printed version would run to more than 250,000 pages. He plans a human-readable write-up in the coming months.
  • Early checks: Scientific American reports that MIT mathematician Andrew Sutherland independently compiled the Lean proof. The Verge noted at publication that the result had not been independently verified and that Anthropic had not commented.
  • Prize rules: the Clay Mathematics Institute offers $1 million for the problem, but its rules require a published solution to survive at least two years of scrutiny and gain general acceptance, The Verge reports.
  • A rival draft: Vakil posted a draft proof of the Generalized Riemann Hypothesis by Alessandro Masullo, reviewed for months by Brian Conrad. About eight hours later, Vakil added that several people had reported problems in that proposed proof, found with help from AI models, and said he makes no claim about its correctness.

Why it matters

Analysis: if it holds up, this would be the clearest sign yet that AI models can contribute to research at the very top of mathematics. The way it was presented matters too. A machine-checked Lean proof shifts much of the review burden to checking that the problem was stated correctly in code, a point Scientific American raises. For now this remains a claim under review, and readers should treat it that way until mathematicians reach a consensus.

Source: The Verge