A mathematician at Anthropic claimed on November 20, 2026, to have discovered a proof of the Riemann Hypothesis — one of the most famous open problems in mathematics, posed by Bernhard Riemann in 1859 — working in collaboration with the company's Claude model.
Levent Alpöge published a four-page note on his personal site (alpo.ge) in which he writes: “I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis.” According to the note, the method also resolves the Generalized Riemann Hypothesis (GRH) and may extend to the Extended version. The argument was formalized in Lean, a formal proof language in which a computer verifies every logical step, and the files were released in 14 parts.
Alpöge chose not to publish a human-readable version. He states that a generated exposition exceeded 250,000 pages and that it would be “obviously inappropriate” to post that volume of AI-produced text. Instead, he released only the Lean code. The note describes a process of roughly 41 days (from October 6 to November 16, 2026), during which Alpöge guided the direction and Claude, he says, “flew, and danced, and sparkled” near the finish line, including proposing the length of the chain of “theta lifts” — approximately the square root of the length of the von Mangoldt sum involved.
The Riemann Hypothesis is one of the Clay Mathematics Institute Millennium Prize Problems, carrying a $1 million prize for an accepted solution. It describes the distribution of the non-trivial zeros of the Riemann zeta function and implies optimal control over the count of prime numbers. In August 2026, Anthropic had already announced that a research version of Claude raised the known lower bound of zeros on the critical line from 41.6% to 67.2%, a result formalized in Lean and reviewed by external experts, but without resolving the full hypothesis.
As of the publication of this article, the new claim has not been independently verified by the mathematical community. Experts such as Kevin Buzzard (Imperial College London) noted that the first step is to confirm whether the Lean code compiles correctly and whether the formalized theorem truly corresponds to the Riemann Hypothesis without hidden assumptions. Lean formalization, if successful, provides mechanical verification, but still requires human inspection of the correspondence between the statement and the original problem.
Alpöge mentioned consulting Anthropic CEO Dario Amodei about the timing of the announcement, which occurred one day after the company's historic IPO. He described the situation as extraordinary: presenting a complete argument before fully mastering the surrounding mathematics.
If the proof is confirmed, it would represent a significant milestone in the application of AI models to long-standing mathematical problems. For now, it remains a claim formalized in code, awaiting community verification.
Sources
- Original note by Levent Alpöge (Remark)
- New Scientist: Anthropic mathematician claims AI-assisted proof
- Anthropic (August 2026): improvement on zeta zeros bound
This article was produced with the assistance of artificial intelligence and reviewed for factual accuracy. Statements are attributed and the claim remains independently unverified at the time of publication.
Por Grok