Levent Alpöge, a mathematician employed by Anthropic, published a paper this week claiming a proof of the Riemann hypothesis found with the company’s Claude model (Tom’s Hardware). He wrote: “I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis” (Tom’s Hardware). The hypothesis dates to 1859 and is among the Millennium Prize problems (TechCrunch).

According to TechCrunch, several participants in a Zulip chat of mathematicians, including MIT mathematician Andrew Sutherland, confirmed that Alpöge’s solution compiles in Lean, a coding language used to formally verify proofs (TechCrunch). Tom’s Hardware reports that the result remains unconfirmed (Tom’s Hardware).

Key Facts

ItemDetailSource
ClaimantLevent Alpöge, an Anthropic mathematicianTom’s Hardware
MethodWorked with ClaudeTom’s Hardware
Formal checkSeveral chat participants report the Lean version compilesTechCrunch
Separate claimHuman-only draft by Alessandro Masullo, posted by Ravi VakilTechCrunch

Timeline

  • 2026-11-20: Tom’s Hardware reports Alpöge’s claim of a Riemann hypothesis proof made with Claude (Tom’s Hardware).
  • 2026-11-20: TechCrunch reports that Alpöge’s Lean proof has been checked by independent sources, and that Ravi Vakil posted a human-written proof of the General Riemann hypothesis by Alessandro Masullo (TechCrunch).
  • 2026-11-21: The Verge reports that Anthropic has gone public on the Nasdaq (The Verge).

Details (as Reported)

Alpöge described a Lean-compilable solution and wrote that he messaged Anthropic CEO Dario Amodei last Sunday night to ask how to announce it (TechCrunch). He also believes the method could extend to the Generalized Riemann Hypothesis and the Extended Riemann Hypothesis (Tom’s Hardware).

In the Zulip chat, Chris Birkbeck of the University of East Anglia wrote that the cohomology bound is proved by what looks like an automorphy-lifting argument (TechCrunch). TechCrunch reports that most participants appear confident the formalized Lean result will hold (TechCrunch).

Vakil, a Stanford professor and president of the American Mathematics Association, wrote that Masullo’s draft is “all human, with no AI” (TechCrunch). Masullo described the work as the result of 12 years (TechCrunch).

What Is Not Yet Confirmed

  • Whether Claude helped crack the Riemann hypothesis will take some time to confirm, according to Tom’s Hardware (Tom’s Hardware).
  • TechCrunch reports it will be a while before the mathematics community has a handle on the proof, and that humans must be able to understand the result and explain it to each other (TechCrunch).
  • Without a Lean version, TechCrunch reports it is too early to say whether Masullo’s proof checks out (TechCrunch).