An Anthropic mathematician claims a Claude-assisted proof of the 167-year-old Riemann hypothesis, which the wider mathematical community hasn’t yet confirmed, though early checks look good.
Levent Alpöge posted the claim on his website on 20 November, as The Verge reported.
In August, Anthropic said a research version of Claude had raised the known share of the Riemann zeta function’s zeros proven to follow the predicted pattern from 41.6% to 67.2%. The company was explicit about the limits of that work: “We don’t expect that the techniques Claude used will lead to proving the Riemann hypothesis.”
In July, Alpöge said he’d solved the Jacobian conjecture using Claude. In September, OpenAI claimed a solution to the Navier–Stokes Millennium Prize problem. Only the Poincaré conjecture has been officially solved in the prize’s history.
I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis.
Levent Alpöge, Anthropic mathematician
Alpöge says he set the direction while Claude did the technical work from 6 October to 16 November. He says the written argument runs to more than 250,000 pages and calls it unreadable. The Lean proof is available on his site in 14 archive files.
He also claims his method resolves the Generalized Riemann Hypothesis and expects it to reach the Extended Riemann Hypothesis with more time. Anthropic hadn’t commented when The Verge reported the claim. Alpöge says he texted CEO Dario Amodei about the result late on a Sunday and indicated he waited until after Anthropic went public on 19 November to publish it.
What does the Riemann hypothesis say?
Prime numbers, such as 2, 3, and 5, are divisible only by 1 and themselves. The Riemann hypothesis, proposed in 1859, predicts a precise pattern in how primes are distributed.
The hypothesis also concerns how the zeros of the Riemann zeta function are distributed.
The hypothesis is one of seven Millennium Prize problems, each worth $1 million from the Clay Mathematics Institute. To qualify, a solution must appear in a qualifying publication, face at least two years of scrutiny, and gain general acceptance among mathematicians. Alpöge has published the proof only on his own site, and the institute hasn’t accepted it.
What have mathematicians checked so far?
Scientific American reports that independent researchers, including MIT mathematician Andrew Sutherland, compiled Alpöge’s proof separately in the Lean proof checker without errors.
Kevin Buzzard of Imperial College London identified the first question to settle: does the statement Lean compiled match the Riemann hypothesis itself? A formal comparison, known as a comparator run, was still pending on Friday.


















