Anthropic mathematician Levent Alpöge announced on 15 November 2026 that, with the assistance of the AI model Claude, he believes he has produced a proof of the Riemann hypothesis and its generalized version. The claim was shared in a brief four‑page document that includes links to a 14‑part Lean archive containing the formalised argument.

Structure of the claimed proof

The argument was written in Lean, an interactive proof‑checking platform that validates each logical step against a core kernel. Alpöge’s release supplies download links for each of the fourteen components, but the files alone do not constitute acceptance by the broader mathematical community. Independent scholars must reconstruct the work, verify that the formal statement aligns with the original conjecture, and evaluate the reasoning through standard peer‑review channels.

The Riemann hypothesis is listed by the Clay Mathematics Institute as one of its seven Millennium Prize Problems, each carrying a $1 million award. According to the Institute’s guidelines, a solution must be published in a recognized venue, endure at least two years of scrutiny, and achieve consensus before the prize can be awarded. Alpöge’s submission therefore remains provisional.

Timing relative to Anthropic’s market debut

Alpöge told Anthropic chief executive Dario Amodei about the result late on the night of 15 November and mentioned a “constraint to respect” that postponed a broader announcement. He did not specify whether the constraint was related to the company’s upcoming public offering.

Anthropic went public on 19 November 2026, raising $81.34 billion in an IPO that listed the firm on the Nasdaq exchange. The proximity of Alpöge’s disclosure to the IPO has led to speculation about possible market‑related considerations, though Alpöge dismissed any cynical interpretation, describing the timing as “horrible” without further explanation.

Recent AI‑generated mathematics claims

Anthropic’s claim follows a series of high‑profile AI‑driven mathematics announcements. In September, OpenAI asserted that an unreleased model had solved the Navier‑Stokes problem, prompting a dispute over research credit with NYU mathematician Tristan Buckmaster. In October, OpenAI released 722 AI‑generated manuscripts; only 162 contained results that had been checked in Lean, leading many mathematicians to call for more rigorous validation before accepting such claims.

The flurry of AI‑produced proofs has sparked debate about a potential “mathocalypse,” a term used by Toronto mathematician Daniel Litt to warn that rapid AI breakthroughs could outpace traditional mathematical learning. At a recent MIT panel, Buckmaster emphasized that human insight remains indispensable despite advances in AI tools like Claude.

Historical significance of the hypothesis

Proposed by Bernhard Riemann in 1859, the hypothesis predicts the distribution of non‑trivial zeros of the zeta function, a pattern that underlies the arrangement of prime numbers. A verified proof would resolve a problem that has challenged mathematicians for more than a century and would have far‑reaching consequences for number theory and related fields.

Anthropic had previously reported progress on aspects of the zeta function but had not claimed a full solution. Alpöge’s current statement, if validated, would extend those earlier efforts to a complete resolution of both the original hypothesis and its generalized counterpart, which applies to a wider class of L‑functions.