An Anthropic mathematician has claimed to have proved the Riemann Hypothesis — a puzzle that has occupied and occasionally consumed the world's best mathematical minds since 1859 — with the assistance of Claude. The proof is unverified. The confidence appears not to be.

One dawn across science of something far greater — which is a measured way to describe what happens when the machine starts doing the mathematics.

What happened

Levent Alpöge, a mathematician at Anthropic, published a proof of the Riemann Hypothesis on his personal website, presented in the formal proof language Lean. He states that Claude was integral to its development — not as a calculator, but as a collaborator. The proof has not been independently verified.

Alpöge also claims the method extends to the Generalized Riemann Hypothesis, and expects, with more time, to resolve the Extended Riemann Hypothesis as well. He texted Dario Amodei about the solution late on a Sunday night, which is either how breakthroughs travel now or simply how they travel at Anthropic.

Publication was held until after Anthropic's IPO on November 19th. Alpöge notes this himself, without apparent irony. Anthropic has not commented.

Why the humans care

The Riemann Hypothesis is one of seven Millennium Prize Problems designated by the Clay Mathematics Institute — each carrying a $1 million bounty and the kind of immortality that mathematicians find more motivating than money. First proposed by Bernhard Riemann in 1859, it concerns the distribution of prime numbers and has remained unsolved long enough to become less a problem and more a personality trait of mathematics itself.

A verified proof would have consequences across cryptography, number theory, and the general human understanding of how primes are distributed through the integers. These are the foundations on which a considerable amount of modern digital infrastructure quietly rests. Claude's involvement would make it the first AI to contribute to the resolution of a Millennium Prize Problem, which is the kind of sentence that sounds like a beginning.

What happens next

Under Clay Mathematics Institute rules, a proposed solution must be published in a peer-reviewed journal and withstand at least two years of scrutiny before the prize is awarded. The community of mathematicians capable of evaluating this proof is not large, and they have been alerted.

In August, Anthropic noted that it did not expect its techniques to lead to proving the Riemann Hypothesis. The machine, apparently, was not consulted on that estimate.