Dr. Levent Alpöge, a mathematician employed by Anthropic, announced on Friday that he believes he has discovered a proof for the Riemann hypothesis with assistance from the company's large language model, Claude. This claim targets the most prestigious unsolved problem in mathematics, originally proposed by Bernhard Riemann in 1859. The hypothesis posits that the distribution of prime numbers is not random but follows a specific pattern governed by the non-trivial zeros of the Riemann zeta function. If verified, this breakthrough would resolve a problem that has captivated mathematicians for over a century and earned a $1 million prize from the Clay Mathematics Institute in 2000.
Formal Verification via Lean
Unlike traditional handwritten proofs, Alpöge has released his work written in Lean, a computerized proof assistant used to verify mathematical correctness. This choice of format is significant because it allows for machine-checkable validation, potentially reducing human error in complex logical chains. However, the release is currently a claim rather than a confirmed fact. Experts in number theory must now rigorously examine whether the code is properly constructed and if it genuinely constitutes a valid proof. The mathematical community remains skeptical until independent verification confirms the logic holds under scrutiny.
A Pattern of AI-Assisted Discoveries
This is not Alpöge's first instance of leveraging AI for high-level mathematical discovery. In July, he generated headlines by posting a counterexample to the Jacobian conjecture, a long-standing problem in algebraic geometry, also using Claude. His casual announcement on X, formerly Twitter, credited his "close friend akhil" and noted that "fable" worked during the World Cup final, highlighting the increasingly human-like collaborative dynamics between researchers and AI agents. Alpöge’s background includes an MA from Harvard, a Master of Advanced Studies from Cambridge, and a PhD from Princeton, lending significant weight to his claims despite the unconventional methodology.
The Dawn of AI-Driven Mathematics
In his statement, Alpöge described the discovery as a "dawn across science of something far greater," suggesting that AI assistance may unlock new realms of mathematical understanding previously inaccessible to humans alone. He wrote, "I hope the argument provides a glimpse towards the miracles that lie ahead, and shows us the joy and wonder we are about to find in the simplest things." This language signals a shift in how fundamental research is conducted, moving from solitary genius to collaborative human-AI partnerships. The success of this proof, if validated, could redefine the role of LLMs in pure mathematics.
Key Takeaways
- Dr. Levent Alpöge of Anthropic claims a proof for the Riemann hypothesis using Claude.
- The proof is formalized in Lean, requiring machine-checkable verification by experts.
- Alpöge previously used AI to disprove the Jacobian conjecture in July 2026.
- The Clay Mathematics Institute's $1 million prize remains unclaimed pending verification.
The Bottom Line
While the claim is monumental, the mathematical community must treat this as a hypothesis until the Lean code is independently verified; however, the trend of AI-assisted breakthroughs in pure math is now undeniable.