Mathematician and AI Both Present Proofs for the Riemann Hypothesis

The Riemann Hypothesis, one of mathematics’ most formidable puzzles since 1859, may finally have found contenders for proof—one driven by a researcher aided by AI, another purely human. The hypothesis, which underpins the distribution of prime numbers and has eluded resolution for over 160 years, was addressed in parallel by Levent Alpöge, a mathematician with ties to Anthropic, and by Alessandro Masullo, a seasoned researcher at Stanford.

AI-assisted proof by Alpöge

Levent Alpöge announced in a recent essay that he’s discovered a proof using the Lean proof assistant, aided heavily by the AI model Claude. In his account, he described working through a long and confusing process involving myriad mathematical concepts—everything from Galois representations to automorphic forms—before arriving at a proof that’s formally checking out in Lean. Despite the clarity in the framework, Alpöge expressed concern that the explanation isn’t yet entirely intelligible to humans, but said he couldn’t wait any longer to publish.

Human proof by Masullo

At the same time, Alessandro Masullo, a postdoctoral researcher, shared a proof of the General Riemann Hypothesis derived from twelve years of independent work. Stanford professor Ravi Vakil highlighted this effort as fully human, with no AI involvement. Masullo’s proof—now in draft form—is intended to be polished and vetted by experts before broader announcement, though the current manuscript has already been released.

What to make of these claims

These simultaneous claims have stirred excitement and caution in the mathematics community. Experts are currently scrutinizing both works—as Alpöge’s Lean-verified result gains traction for its formal structure, and Masullo’s traditional mathematical proof is evaluated for originality and rigor. The community is particularly focused on whether Alpöge’s explanation can be made fully human-understandable, and whether Masullo’s draft holds up under peer review.

These developments arrive after a summer marked by intense efforts from mathematicians and frontier AI labs alike, competing to resolve longstanding mathematical challenges. Alpöge’s announcement coincides with Anthropic going public, adding extra weight to his AI-driven proof. Both achievements make clear that AI is becoming a serious partner in foundational research—but human creativity and persistence remain essential.

It’s too early to declare the Riemann Hypothesis solved. What matters now is rigorous peer review, clarity of exposition, and community consensus. These proofs, AI-assisted or not, mark a potential turning point in how we approach deep mathematical problems—and how AI could reshape proof in fundamental areas.

What this means: If any of these proofs stand up under scrutiny, it would mark the end of a 167-year mystery and redefine the frontier between human and AI in mathematics. The key question now is which version—AI-assisted or human-penned—will earn the full trust of mathematicians worldwide, and how this moment transforms both math and AI research going forward.