Anthropic mathematician claims Riemann hypothesis proof with Claude
1859 prime-number puzzle carries million-dollar Clay prize, claim lands days after employer's record IPO and awaits Lean verification
Images
Dr Levent Alpöge, a mathematician at Anthropic, announced on Friday that he believes he has, working with the AI model Claude, discovered a proof of the Riemann hypothesis. The conjecture, proposed by Bernhard Riemann in 1859, holds that the positions of prime numbers along the number line follow a meaningful order rather than randomness, and the Clay Mathematics Institute has offered one million dollars for its proof since 2000. Alpöge posted the claim on his personal website together with a version of the argument written in Lean, a programming language that verifies the correctness of mathematical proofs line by line.
The hypothesis was one of David Hilbert's 23 challenges for the twentieth century, and claimed proofs have surfaced regularly ever since — none has survived expert scrutiny. Two things set this attempt apart. The Lean version turns verification from a question of authority into a question of computation: experts must now establish whether the code is properly written and whether it proves what it claims. And the announcement lands days after Anthropic completed the largest stock-market debut on record, from a company whose product is the very model now credited with historic mathematics.
Alpöge is not a crank. He holds a master's degree from Harvard, a Master of Advanced Studies from Cambridge and a doctorate from Princeton, and citizenship of both the United States and Turkey. In July he posted a counterexample to the Jacobian conjecture, a well-known problem in algebraic geometry, found — like the Riemann work — with Claude; that discovery he announced on X with the words "hello there the jacobian conjecture is false."
How the labour was divided between mathematician and model remains undefined; the statement says "working with Claude" and nothing more precise. Nor would a genuine proof have the consequences periodically feared in popular accounts: internet encryption rests on the difficulty of factoring specific large numbers, not on ignorance of how primes are distributed, and a proof would settle the foundations of that distribution rather than shake them. Alpöge himself aims higher, writing that what he saw during the work "so amazed me that I believe it is one dawn across science of something far greater."
Until that checking is done, the most famous unsolved problem in mathematics stands addressed by a statement on one mathematician's website and a Lean file awaiting its first readers.