Anthropic Mathematician Claims Claude Helped Prove the Riemann Hypothesis
Levent Alpöge has published a proposed proof of the 167-year-old Riemann Hypothesis written in Lean with assistance from Anthropic's Claude model.
Levent Alpöge, a mathematician working at Anthropic, has published a proposed proof of the Riemann Hypothesis, widely regarded as one of the most critical open problems in number theory. First proposed by Bernhard Riemann in 1859, the 167-year-old challenge concerns the distribution of prime numbers and the zeros of the Riemann zeta function. Alpöge claims he formulated the solution alongside Anthropic's Claude LLM, writing the formal arguments in the Lean proof assistant and releasing the work on his personal website shortly after Anthropic floated its initial public offering on the Nasdaq.
According to source reports, Alpöge's announcement asserts that the core methodology developed with Claude extends beyond the initial formulation to address both the Generalized Riemann Hypothesis and the Extended Riemann Hypothesis. Alpöge stated on his website that he and Claude formalized the arguments without relying on unproven assumptions. He noted that he messaged Anthropic Chief Executive Officer Dario Amodei about the proof late on a Sunday night, deliberately holding off on public dissemination until after the company completed its stock market listing later that week.
Anthropic has not officially commented on or involved itself with Alpöge's published findings, and the company had previously expressed skepticism regarding automated breakthroughs on the problem. In August, Anthropic confirmed it was utilizing Claude to explore the Riemann Hypothesis but noted at the time that it did not expect those specific techniques to yield a complete proof. Furthermore, the Clay Mathematics Institute, which oversees the Millennium Prize and offers a $1 million reward for verified solutions, requires proposed proofs to undergo at least two years of formal peer scrutiny and achieve broad community consensus before granting official recognition.
The development highlights the growing role of large language models and interactive theorem provers in advanced mathematical research. Alpöge previously claimed to have resolved the Jacobian conjecture using Claude, underscoring an accelerating shift toward machine-assisted formalization. The Riemann Hypothesis, which asserts that all non-trivial solutions of the zeta function equation lie on a specific critical line, had previously been computationally verified for its first 10 trillion solutions, but an overarching formal proof had eluded human mathematicians for over a century.
This latest claim follows a turbulent period for AI-driven mathematical research. Earlier in the year, OpenAI stated it had cracked the Navier-Stokes problem—another Millennium Prize challenge governing fluid dynamics—by deploying 10,000 parallel automated agents utilizing Codex and Claude. However, OpenAI's attempt faced significant backlash over research practices and allegations that it relied on unpublished human work from collaborators, including Alpöge and New York University professor Tristan Buckmaster. While OpenAI denied wrongdoing, the controversy underscored the tension surrounding attribution and validation in AI-generated mathematical proofs.
As mathematicians begin examining Alpöge's Lean code, the global research community faces the arduous task of verifying the complex logical chains generated by the human-AI collaboration. The Clay Mathematics Institute has officially recognized only one Millennium Prize problem solution in twenty-five years—Grigoriy Perelman's proof of the Poincaré conjecture. If Alpöge's formalization holds up under rigorous peer review, it will mark a major validation for automated reasoning tools in formal mathematics.