Levent Alpöge, a mathematician on the technical staff at AI research firm Anthropic, announced Friday that he used a non-public version of the company’s Claude chatbot to prove the Riemann Hypothesis, widely viewed as the world’s most famous unsolved mathematics problem.
Alpöge published his work on his website, stating that he spent about six weeks guiding Claude through the reasoning required to crack the 167-year-old puzzle. First formulated in 1859 by German mathematician Bernhard Riemann, the hypothesis proposes a specific pattern for how prime numbers appear. Proving the conjecture is classified as a Millennium Prize Problem by the Clay Mathematics Institute, which offers a $1 million bounty for a verified solution.
A Quarter-Million Pages of Code
The proposed proof spans more than 250,000 pages written in Lean, a computer language designed to formalize and check mathematical logic, according to reports by Business Insider and the BBC. Alpöge stated on his website that the technique also extends to the Generalized Riemann Hypothesis. He described Claude’s performance near the end of the process as stepping up to complete complex steps in the argument.
According to Alpöge, he completed the breakthrough on Sunday but delayed announcing his findings until after consulting with Anthropic Chief Executive Officer Dario Amodei. The timing coincided with Anthropic’s initial public stock offering on Thursday. Alpöge addressed concerns over the timing in a public statement, taking personal responsibility for the release date and denying that the announcement was intended as a publicity tactic.
Verification and Industry Reaction
Mathematicians emphasized that the claim will require extensive review before it can be validated. Steven Strogatz, a mathematics professor at Cornell University, told Business Insider that while a computer-based system had checked Alpöge’s solution, automated systems can make errors and the proof remains unverified by independent experts. Strogatz called the Riemann Hypothesis “the Moby Dick of Mathematics.”
Terence Tao, a prominent mathematician at UCLA, posted on his website that digesting the massive document “will take some time,” though he noted that having a human author available to clarify details will assist experts during the review. To qualify for the prize, rules established by the Clay Mathematics Institute require a proposed solution to be published and endure at least two years of review across the global mathematical community.
Rising Role of AI in Complex Math
Alpöge’s announcement comes amid a surge of high-profile mathematical claims powered by artificial intelligence models. In September, rival AI company OpenAI claimed its systems solved part of the Navier-Stokes equations, another Millennium Prize Problem regarding fluid movement. OpenAI later announced progress or solutions for hundreds of other outstanding math problems, prompting debate within academic circles over the impact of AI on mathematical research.
Earlier this year, a group of mathematics scholars published an open letter warning that AI companies boasting of major discoveries to market commercial software could prove “detrimental to the science of mathematics,” the BBC reported.
What happens next depends on the global mathematical community’s ability to inspect and verify Alpöge’s quarter-million-page document. Anthropic has not issued an official corporate statement regarding the claim, and independent scholars expect the evaluation process to take months or years.



