An Anthropic mathematician named Levent Alpöge claims he has, working with Claude AI, discovered a proof of the Riemann Hypothesis, a 167-year-old mathematical puzzle worth a $1 million prize [1]. The claim surfaced on Friday, November 21, 2024, one day after Anthropic's stock-market debut on Nasdaq on Thursday, November 19, where the company raised $81.34 billion in its offering [1]. Alpöge wrote in a four-page note that the method also resolves the broader Generalized Riemann Hypothesis, and says he and Claude encoded their argument in Lean, software used to verify mathematical proofs step by step, with 14 downloadable parts of the proof archive available [1]. However, the claim has not yet been independently verified by the mathematical community, and availability of the proof does not amount to acceptance [1].
The timing of the announcement has drawn scrutiny. Alpöge acknowledged messaging Anthropic CEO Dario Amodei about a possible announcement "late Sunday night" on November 15, and subsequently consulted company leadership, noting there was "a constraint to respect" without explaining precisely what it was [1]. He stated "the precise timing is horrible" and rejected a cynical interpretation of the timing relative to the IPO, but did not explicitly deny that the company's public offering influenced the announcement delay [1]. This context is significant given recent skepticism in the mathematical community about AI-generated proofs: OpenAI claimed in September to have solved the Navier-Stokes Millennium Prize problem, sparking a credit dispute with researcher Levent Alpöge and NYU mathematician Tristan Buckmaster over research priority [1]. OpenAI subsequently released 722 manuscripts in October from an unreleased model, but only 162 included Lean-checked main results, leading mathematicians to demand stronger evidence rather than accepting the company's flood of claims [1].
The broader context reveals growing concerns about AI's role in mathematics. Some mathematicians fear a "mathocalypse," with AI solving coveted problems before researchers can learn from pursuing them, with Toronto mathematician Daniel Litt warning that without adaptation, "there's just no more math in 50 years" [1]. At an MIT panel this week, mathematician Tristan Buckmaster argued that mathematics still requires human understanding [1]. The Riemann Hypothesis, first posed by Bernhard Riemann in 1859, predicts where certain zeros of the Riemann zeta function fall and matters because those zeros are linked to the distribution of prime numbers, whose patterns underlie extensive mathematical research [1]. A verified proof would represent a historic mathematical result after over a century of attempts [1].
CONCLUSION
Anthropic's claim of an AI-assisted proof of the Riemann Hypothesis, announced one day after the company's $81.34 billion IPO, has generated significant attention but remains unverified by the mathematical community [1]. The suspicious timing and recent skepticism toward AI mathematical claims—including OpenAI's disputed Navier-Stokes solution—suggest the market should await independent verification before assessing the credibility of this breakthrough [1]. The announcement reflects both the potential and the credibility challenges facing AI companies in high-stakes scientific claims.