Anthropic mathematician claims Riemann Hypothesis proof with Claude; Lean community certifies it correct
Anthropic mathematician Levent Alpöge said on November 20 that he and Claude found a proof of the Riemann Hypothesis, published as a formal proof in Lean. On November 21, the Lean and Mathlib communities issued a joint statement certifying the proof as correct. No human-readable paper exists yet.
Levent Alpöge, a mathematician at Anthropic, said on November 20 (local time) that he and the company's AI model Claude have found a proof of the Riemann Hypothesis. On November 21, the Lean and Mathlib communities issued a joint statement certifying the proof as correct. The report comes from Japanese-language outlet ITmedia, which cites Scientific American for the outside verification.
What was announced
Alpöge published his statement on his own website. The proof is released as a formal proof written in the Lean proof assistant. He also says the method resolves the Generalized Riemann Hypothesis, a broad extension of the original problem.
He describes the collaboration this way: near the finish, "Claude far surpassed me." The statement includes a timeline compiled by Claude. According to that timeline, key ideas for the proof emerged one after another from October 6, and the proof of the Generalized Riemann Hypothesis was reached on November 16.
The Riemann Hypothesis was proposed by Bernhard Riemann in 1859. It states that all non-trivial zeros of the Riemann zeta function lie on a single line. It is one of the Clay Mathematics Institute's Millennium Prize Problems, which carry a $1 million prize. The Poincaré conjecture is the only one solved so far.
Verification
Lean mechanically checks that each step of a proof is logically valid. According to Scientific American, outside researchers, including MIT mathematician Andrew Sutherland, compiled the published code independently and it completed without errors. Sutherland said the chance that something is wrong is "almost zero."
The joint statement from the Lean and Mathlib communities is backed by Lean FRO, the organization developing Lean, among others. It reports the following:
- The proof runs to 1700万 lines of Lean (the source's figure, using the Japanese counting unit 万).
- Dozens of people reproduced the compilation.
- Two independent tools confirmed that the theorem proved matches the statement of the Riemann Hypothesis in Mathlib, Lean's mathematics library.
- Multiple independent checkers also inspected the proof.
- The check used four criteria for evaluating Lean proofs, published on November 12 on the blog of Kevin Buzzard of Imperial College London, and the proof met all of them.
The statement adds that the work of having mathematicians understand and explain the proof's new ideas is still ahead.
No readable paper yet
Alpöge has not published a paper that mathematicians can read. He says the proof is extremely long. A detailed document generated from the formal proof exceeded 25万 pages when he tried it, which he called unreadable. He released only the Lean code and plans to write a formal paper once he understands the content.
Earlier AI results
In August, Anthropic announced that an unreleased Claude had produced a related result. It raised the lower bound on the proportion of zeros lying on the line from 41.6% to 67.2%, which was not a proof of the hypothesis itself. In October, OpenAI published a batch of math results from an internal AI, including a proof of a weakened form called the "quasi-Riemann Hypothesis." The current claim differs because it asserts a proof of the Riemann Hypothesis itself.
Scientific American also noted that the announcement came the day after Anthropic's stock listing. Alpöge acknowledged in his statement that the timing could invite cynical readings and said such readings are mistaken.
What this means
The claim rests on a machine-checked artifact rather than on a human-readable argument. The Lean and Mathlib communities certified the formal proof, and external researchers reproduced the compilation. The certification covers logical validity and the match with Mathlib's statement of the hypothesis. The communities say the human work of understanding and explaining the proof's ideas has not been done. Until Alpöge's planned paper appears, the proof exists as Lean code that mathematicians cannot yet read in conventional form.
Related Articles
Anthropic mathematician Levent Alpöge claims Claude-assisted proof of the Riemann hypothesis, posts it in Lean
Levent Alpöge, a mathematician employed by Anthropic, says he has proven the Riemann hypothesis, first proposed in 1859, working with Claude. He posted the proof coded in Lean, and experts have yet to confirm that it is correct.
AWS releases I/Q, I/R and Isabelle Assistant to connect AI agents to the Isabelle proof checker via MCP
AWS has made three tools in its AutoCorrode verification suite publicly available: I/Q, I/R and Isabelle Assistant. They connect AI agents to the Isabelle proof assistant through the Model Context Protocol, so generated proofs are checked by computer. Amazon Science authors demonstrate the setup using Claude Opus 4.6.
Anthropic Adds Experimental Code Mode to Claude Code v2.1.323, Chaining Tool Calls in JavaScript
Anthropic has added an experimental Code Mode to Claude Code v2.1.323. Claude can write JavaScript that chains Bash, Read and Edit calls in one step, and keep talking while long commands run.
Anthropic launches Linked billing beta, letting Claude API developers bill usage to a customer's own organization
Anthropic launched Linked billing in beta on November 17, 2026. Developers can bill Claude API usage in their applications to a customer's own Claude organization, choosing per request whether the customer or the developer pays.
Comments
Loading...