Anthropic · 2026-11-20 · major
Riemann Hypothesis proof claim — Anthropic's Levent Alpöge and Claude
In short
Anthropic mathematician Levent Alpöge says he and Claude found a proof of the Riemann Hypothesis and formalized it in Lean. He also claims the Generalized Riemann Hypothesis. No outside expert has verified it yet.
Anthropic · 2026-11-20 · 3 sources checked

An Anthropic mathematician says Claude helped him prove math's most famous open problem; the Lean files are public, unchecked.
Quick facts
| Claimed result | Riemann Hypothesis + Generalized RH |
|---|---|
| Who | Levent Alpöge (Anthropic), working with Claude |
| Formal proof | Lean, 14-part archive on alpo.ge |
| Written paper | Not yet; Alpöge says he will write it properly |
| Status | Not independently verified (Nov 20, 2026) |
What is it?
Levent Alpöge, a mathematician at Anthropic, posted a short statement on November 20, 2026 saying: "I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis." The Riemann Hypothesis is the classic open problem about how prime numbers are spread out. Alpöge says the same method also settles the Generalized Riemann Hypothesis, and he expects the Extended Riemann Hypothesis to follow with more time.
How does it work?
According to Alpöge's statement, the argument descends from Stepanov's method, which proves the Riemann Hypothesis for function fields using a divisor with multiplicity. Over the integers he and Claude used chains of theta lifts to mimic that multiplicity, and Claude suggested a chain length of √X for a von Mangoldt sum of length X. The proof is written in the Lean proof assistant, released as a 14-part archive on alpo.ge.
Why does it matter?
A Lean formalization can be checked by machine, so mathematicians can test whether the formal statement really encodes Riemann's question and whether the proof compiles. The problem has stood since 1859, and a working proof found with an AI model would change how mathematicians use these tools. Scientific American reports that no expert has verified it yet.
Who is it for?
mathematicians, formal-verification and AI-for-math researchers
Frequently asked questions
- Has the Claude Riemann Hypothesis proof been verified?
- No. Scientific American reported on November 20, 2026 that experts have not yet confirmed Levent Alpöge's Lean code or the proof. Reviewers need to check two things: that the Lean statement truly encodes the Riemann Hypothesis, and that the proof is correct. Alpöge's own statement calls the argument incredibly long at the moment and says he expects it to be simplified.
- Why does a Lean formalization matter for a Riemann Hypothesis claim?
- Lean is a proof assistant that checks every logical step by machine, so a formal Riemann Hypothesis proof can be tested without trusting the author's prose. Alpöge says he and Claude formalized the arguments without assumptions. The remaining human work is confirming that the formal statement matches Riemann's question and that the proof holds, which outside mathematicians have not yet done.
- Is there a readable paper for Alpöge's Riemann Hypothesis proof?
- Not yet. Levent Alpöge's statement says it is routine to produce a huge, unreadable document with full detail, but that he intends to understand the argument, simplify it, speak about it and write it properly. For now the public material is a one-page statement and the 14-part Lean archive hosted on his site, alpo.ge.
- How is this different from Claude's earlier Riemann zeta result?
- Claude's earlier result, already in the AI/TLDR feed, raised a known bound on Riemann zeta zeros, which is partial progress. Levent Alpöge's November 20 claim is a full proof of the Riemann Hypothesis itself, plus the Generalized Riemann Hypothesis, built around chains of theta lifts. It is a much larger claim and is still unverified by outside experts.
Try it
Read the statement at https://alpo.ge/remark.pdf; the Lean archive parts are linked at its footnote.