- Alpöge delivered the proof as ~1.3GB of Lean code formalizing GRH, with experts estimating under 1% chance of a Lean bug
- Masullo's 940-page human proof underwent years of secret review, offering a direct traditional-math counterpoint
- The episode shows AI-generated proofs hinge on formal verification toolchains rather than human read-through
AI CURATED HOT STORY · 11/20/2026
Anthropic Mathematician Claims Riemann Hypothesis Proof with Claude
Anthropic mathematician Levent Alpöge has publicly claimed that, in collaboration with Claude, he proved the 167-year-old Riemann Hypothesis, and further asserted that the method extends to the Generalized Riemann Hypothesis. According to his account, Claude proposed key proof steps in advanced areas beyond the mathematician's own expertise, and over roughly six weeks they formalized the complete argument into machine-verifiable Lean code, allegedly without additional assumptions. If confirmed by peer review, this would be the first substantive AI contribution to a top-tier mathematical problem, marking a major leap in AI mathematical capabilities.
Currently, the claim is only Alpöge's personal public assertion and has not undergone peer review or independent verification by the mathematical community. Anthropic has not participated in or commented on the claim; Alpöge said he privately informed CEO Dario Amodei and delayed the announcement until after the company's IPO, raising concerns about conflicts of interest. Multiple sources consistently emphasize that the proof was published as Lean code, but formal verification itself still requires community review, and OpenAI's mathematical results announced in October are believed to have provided a breakthrough.
If ultimately confirmed, this event would profoundly impact the ecosystem of AI-assisted mathematical research, demonstrating the potential of large models in the highest-order creative mathematical reasoning and promoting the adoption of formal verification and AI collaboration paradigms. However, until independent verification is complete, the industry should remain cautious and avoid overinterpreting personal claims.
Hourly heat curve
20 fully observed hoursThe line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.
Coverage timeline
Total 10 reports · Latest first- Latent SpaceT1·88 pts
- Rohan Paul (@rohanpaul_ai)T2·88 pts
- Claude contributed a key step in the Riemann hypothesis proof, in advanced areas the human collaborator had not mastered
- The proof took about six weeks and was formalized as machine-checkable code, making it verifiable
- The event validates Peter Thiel's 2024 prediction that AI would rapidly conquer hard math problems
- IT之家 · IT业界T2·88 pts
- Alpöge says Claude's late-stage contribution exceeded his own, but only a 4-page note and 14 Lean archives were released
- Tao notes the proof appears to rely on the Langlands program and arithmetic geometry, requiring time for the community to verify
- Preliminary checks confirm the formalized statement is GRH, though past Lean versions had kernel defects
- Hacker News AIT2·88 pts
- Alpöge reports Anthropic models produced a Lean formalization of the Generalized Riemann Hypothesis proof spanning 17.4M lines
- It follows an AI-assisted result bounding nontrivial zeros within 3/8 of the critical line, signaling accelerating AI-for-math progress
- The proof has not yet been independently verified, so credibility awaits community review
- Artificial Analysis (@ArtificialAnlys)T1·71 pts
- Levent Alpöge claims a Riemann hypothesis proof produced with Claude, not yet peer-verified
- OpenAI said in September an unreleased internal model solved Navier-Stokes, also unverified
- Only the Poincaré conjecture is officially recognized among the seven Millennium Problems
- TechCrunch AIT2·88 pts
- Anthropic mathematician Levent Alpöge claims a Lean-compilable proof of the Riemann Hypothesis, open since 1859
- Lean community mathematicians are actively scrutinizing the proof while a second researcher claims an independent one
- The episode highlights the growing intersection of AI labs and formal math verification tooling
- Rohan Paul (@rohanpaul_ai)T2·88 pts
- Claude proposed a key proof step in advanced areas the mathematician had not mastered
- The full argument was formalized as machine-checkable code over roughly six weeks
- Currently a personal claim awaiting peer review by the mathematics community
- The Verge AIT2·78 pts
- Alpöge claims Claude helped prove the Riemann hypothesis, one of seven Millennium Prize problems
- The proof is unverified; Anthropic neither participated nor commented
- Alpöge says he privately told the CEO and delayed publication until after the IPO, raising conflict-of-interest questions
- Simon Willison 博客T2·88 pts
- Alpöge claims a Riemann Hypothesis proof co-discovered with Claude, extensible to the Generalized Riemann Hypothesis
- He says the arguments were formalized without additional assumptions
- The claim is a personal public statement, not yet peer-reviewed or independently verified
- IT之家 · 智能时代T2·88 pts
- Alpoge claims the proof covers not only Riemann but the broader Generalized Riemann Hypothesis
- The proof is published as Lean code and awaits formal verification
- OpenAI's October math results are credited with opening the breakthrough path