Anthropic staff member says AI helped solve Riemann Hypothesis
Levent Alpöge claims a proof of the 1859 math problem using Claude, released just after the company's record-setting initial public offering.
By Emily Lawson ·

More than 250,000 pages of text emerged from a collaboration between an Anthropic employee and the company's AI model, a volume of work that culminated in a claimed solution to one of mathematics' oldest unsolved problems.
Levent Alpöge, a member of Anthropic's technical staff, announced the discovery on Friday. He stated that he believes he has proven the Riemann Hypothesis with the assistance of Claude. The problem, first formulated in 1859, concerns the distribution of prime numbers as they approach infinity. It has remained unsolved for over 150 years.
Alpöge said the breakthrough occurred late on Sunday night. At that time, he messaged Anthropic CEO Dario Amodei to ask about his ability to make the announcement public. The timing was tight because the San Francisco-based company was preparing to make major news with its stock market debut. The company completed its initial public offering on Tuesday, raising a record $81 billion at a $1.5 trillion valuation. Alpöge revealed his findings the day after that financial milestone.
The proof was released as 17 million lines of code in Lean, a programming language designed for verifying mathematical theorems. Alex Kontorovich, a distinguished professor of mathematics at Rutgers University, noted that such findings are typically published as formal papers rather than code. He explained that traditional papers require readers to manually verify each line to build understanding. In this case, Claude followed the code and confirmed the proof is true. Kontorovich said he does not yet fully understand the result, but he trusts the computer's strictness. He noted that if the code satisfies the machine, there is a very good chance it will satisfy human scrutiny as well.
Alpöge acknowledged that he is presenting the argument before mastering the surrounding mathematics. He wrote on his website that Claude far outpaced him near the finish line. He described the AI's performance as flying and sparkling during the final stages of the work. Anthropic did not respond to a request for comment regarding the claim.
This marks the second time in recent months that an AI-linked entity has claimed to crack a famous math problem. In September, OpenAI stated that its technology was key to a breakthrough with Navier-Stokes equations, which govern fluid flow. Kontorovich expressed surprise at the rapid sequence of events. He said he feels an out-of-body experience while trying to comprehend the claim that the Riemann Hypothesis has been solved.
The Clay Mathematics Institute, which selected the Riemann Hypothesis in 2000 as one of seven Millennium Prize Problems, has not yet updated its status. The problem remains listed as unsolved on the institute's website. Martin Bridson, the institute's president, said in a statement that the announcement holds historic significance. He noted that the journey from this announcement to a place of real understanding will be long and difficult. The institute offers a $1 million reward for a verified solution.
Anthropic raised its record $81 billion this week through its initial public offering. CEO Dario Amodei has projected that AI will lead to major advancements in math and science while also posing risks. The company's valuation reached $1.5 trillion during the offering. Alpöge's four-page explanation is currently posted on his personal website.






