Skip to content
VECTOR WIREAI INTELLIGENCE
UTC
Home

Anthropic mathematician claims Claude-assisted proof of Riemann hypothesis

Lean-formalized proof spans 14 compressed files; no readable paper exists and the informal draft exceeds 250,000 pages.

Levent Alpöge, a mathematician employed by Anthropic, claimed on November 20 to have proved the Riemann hypothesis — the most famous open problem in mathematics — working with Claude, and posted the proof formalized in the Lean programming language1,13. The solution has not been independently verified12. Separately, Alpöge and mathematician Tristan Buckmaster published a paper the same day claiming that smooth initial conditions can make the unforced three-dimensional Navier-Stokes equations blow up in finite time, which would settle a second famous open problem9.

"I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis," Alpöge wrote in a statement on his website6. "In fact, the method easily extends much further, resolving the so-called Generalized Riemann Hypothesis, and I expect, with more time, the Extended Riemann Hypothesis too".

A 167-year-old conjecture

First proposed in 1859 by Bernhard Riemann, the hypothesis asserts that the locations of prime numbers along the number line follow a meaningful order rather than appearing randomly. It is one of seven Millennium Prize problems, widely considered among the most formidable open challenges in mathematics. Scientific American called it "the North Star for the entire field of number theory" and noted that progress toward solving it was so rare that most top experts spent little time working on it.

Alpöge posted a four-page note on his website and a Lean formalization split across 14 compressed files. He described a route inspired by Stepanov's proof of the Riemann hypothesis over function fields and said Claude suggested the decisive step: a long chain of theta lifts, with the chain length set at roughly the square root of the size of the primes involved. The note says no readable paper exists and that the informal draft runs past 250,000 pages. Of Claude's role, Alpöge said: "Claude far outpaced me near the finish line. No, Claude flew, and danced, and sparkled near that finish line".

Number theorist Chris Birbeck of the University of East Anglia was among mathematicians examining the result on the Zulip chat platform, noting that the cohomology bound uses an automorphy-lifting argument drawing on Galois representations, base change, and other advanced techniques2.

A second claim: Navier-Stokes blow-up

The Navier-Stokes paper, co-authored with Buckmaster, runs 79 pages and is accompanied by a GitHub repository, but the numerical certificate that verifies the result is still missing. Buckmaster wrote on Mastodon that the write-up is "unpolished and heavily LLM-assisted". Scott Armstrong, a mathematician at New York University and Sorbonne, wrote on X that the pair "claim to have proved blow-up of UNFORCED Navier-Stokes," a post that drew about 33,000 views within hours.

Anthropic's distance, and the IPO backdrop

Anthropic has not commented on or involved itself in Alpöge's announcement. Alpöge said he texted Anthropic CEO Dario Amodei about the solution "late Sunday night" and, according to The Verge, appeared to have held off publication until after the company's IPO on November 19. Anthropic began trading on Nasdaq that day at $111.50 per share, 34% above its $83 IPO price[2]. The company's public offering raised over $93 billion5.

AI's accelerating role in mathematics

The claim arrives amid a rapid sequence of AI-driven mathematical results. On October 6, OpenAI posted over 700 AI-written math manuscripts, roughly 300 of which included Lean proofs10. One of those manuscripts claimed a zero-free region for the Riemann zeta function where the real part exceeds 7/8. Months before that, an unreleased Claude model raised the proven share of zeta zeros on the critical line from 41.6% to 67.2%, with a Lean proof.

Not all mathematicians welcome the pace. In September, Terence Tao and 24 other Fields Medalists signed a declaration stating that "the goals of the AI companies and the goals of the mathematical community are severely misaligned". At a lecture at Caltech on October 9, Tao said: "Further blind optimization of problem-solving alone is now actively harmful to the long-term health of mathematics".

ANALYSIS If both the Riemann and Navier-Stokes claims withstand independent verification, they would represent the highest-profile mathematical results yet attributed to AI-assisted work, settling two problems that have stood open for more than a century and six decades respectively.

Was this useful?

The day’s vector, in your inbox.

Vector Wire Daily: one dispatch each publishing day.