Anthropic mathematician claims AI was used to prove Riemann hypothesis
Anthropic mathematician Levent Alpöge announced on Friday that he spent about six weeks using a non-public version of the company's Claude chatbot to produce a detailed proof of the Riemann hypothesis. Named after Bernhard Riemann, the nearly 170-year-old problem—widely regarded as the single most famous in mathematics—proposes that prime numbers have a determined order along the number line. Alpöge's proof, written in the Lean verification language, spans more than 250,000 lines.
The announcement came a day after Anthropic's US stock market debut. Alpöge called the timing "horrible" while rejecting cynical interpretations, noting he had messaged CEO Dario Amodei on Sunday to ask how to announce the discovery. Prominent mathematician Terrance Tao said on Friday that verifying the proof remains some way off, though having a human author to consult will help the digestion process.
The claim follows recent OpenAI announcements in higher mathematics, including a September claim that over 10,000 mostly autonomous AI agents solved part of the Navier-Stokes fluid equations in 88 hours, and a release a month later covering 372 other unsolved problems. Those results remain unverified, and a group of scholars has warned that firms boasting about solutions for commercial showcase is "detrimental to the science of mathematics."
Continue in the app — vote & join in ➔