The Lean and Mathlib communities say they have verified a Claude-assisted proof of the Riemann Hypothesis submitted by Anthropic mathematician Levent Alpöge. In a joint statement published on November 21, the groups said they had full confidence in the proof and confirmed that it establishes the Riemann Hypothesis statement already formalized in Mathlib, the mathematical library used by Lean.
Alpöge released the proof on November 20 as a Lean library containing about 17 million lines of code. Lean is a programming language that allows mathematicians to express theorems in a form that software can check step by step. The Lean and Mathlib communities said the code compiled without errors, the theorem matched Mathlib’s reviewed definition of the Riemann Hypothesis, and the exported proof was accepted by Lean’s kernel and several independent checkers, including con-leche and con-ron.
Alpöge said he worked with Anthropic’s Claude for about six weeks and that the model helped develop the argument. In a note accompanying the formal proof, he described the argument as extremely long and said he intended to prepare a fuller explanation that mathematicians could read more easily. He also said the method extends to the Generalized Riemann Hypothesis and that he expects it may extend further.
A separate verification page maintained by MIT mathematician Andrew Sutherland, updated on November 22, reports that all 175,187 Lean modules compiled without errors and passed a kernel replay. It also records successful checks by independently written proof checkers, including nanoda and con-leche. Two other end-to-end checks were still listed as running in the page’s latest update.
The Clay Mathematics Institute still lists the Riemann Hypothesis as unsolved. Its prize rules require a proposed solution to appear in a qualifying publication, remain available for at least two years and gain general acceptance in the mathematics community before the institute considers a prize claim. The Associated Press reported that Rutgers mathematician Alex Kontorovich had not yet fully understood the proof, while institute president Martin Bridson described the path from the announcement to broad mathematical understanding as long and difficult.
The development follows Abijita’s earlier report on OpenAI’s AI-assisted work on the Navier-Stokes problem, another Millennium Prize challenge. Anthropic has also described how Claude helped researchers investigate the origins of NSA-supplied elliptic-curve seeds, a separate mathematical and cryptographic question.



