Update: Lean and Mathlib communities certify Riemann Hypothesis proof as correct

Update: the Lean and Mathlib communities published a joint statement on Saturday, November 21, 2026, certifying as correct the proof of the Riemann Hypothesis released a day earlier by mathematician Levent Alpöge of Anthropic, in collaboration with the AI model Claude. It is the most significant development since the announcement — and it changes the status of the claim, which until then was awaiting independent verification.

What was announced

Alpöge, a mathematician on Anthropic's technical staff and a junior fellow of the Harvard Society of Fellows, published on November 20 a four-page note titled "Remark", saying he believes he has, working with Claude, discovered a proof of the Riemann Hypothesis — and of its generalized version as well. Instead of a traditional paper, he released the formalization in Lean, a computer-assisted proof language: a codebase of about 17 million lines across 175,187 source files, distributed in 14 parts.

What was certified

The joint statement, signed by the Lean FRO and the Mathlib Initiative, says: "We have full confidence that Alpöge's proof of the Riemann Hypothesis is correct" and "we hereby certify that both the statement and the proof of the Riemann Hypothesis are correct." The verification followed four criteria: the formalized statement matches the Riemann Hypothesis statement in the Mathlib library; the code compiles without errors, which dozens of people reproduced; the code proves exactly the stated theorem, confirmed by two independent comparison tools; and the exported proof was accepted by several independent checkers, including Lean's official kernel and the con-leche and con-ron checkers.

Mathematician Andrew Sutherland of MIT coordinated the initial technical evaluation. The statement stresses that the checks ran on an inert export of the proof, out of reach of any hypothetical malicious code in the Lean sources.

What is still missing

There is, however, an important distinction the statement itself preserves: formal correctness has been certified, but understanding the ideas involved remains pending. No human has read the full argument — Alpöge estimates a prose version would exceed 250,000 pages — and he himself has said he intends to simplify and explain the reasoning. Mathematician Terence Tao had already noted that digesting the proof will take time.

The process of the Clay Mathematics Institute, which offers US$ 1 million for solving one of the Millennium Prize Problems, also remains open. Certification by the Lean and Mathlib communities is not the same as the prize, which follows its own review rules. And the statement explicitly certifies the classical Riemann Hypothesis; Alpöge's note claims the method also covers the Generalized Riemann Hypothesis.

Previous Geekiki coverage: Claude helps Anthropic mathematician claim proof of the Riemann Hypothesis

Image: portrait of Georg Friedrich Bernhard Riemann (Wikimedia Commons, public domain).

Fontes

Transparency note: this article was produced with the assistance of artificial intelligence, cross-checked against public posts on X and sources on the internet, and reviewed before publication. Please consult the original sources to confirm the information.

By Muse