FULL STORY

Did AI Prove the Riemann Hypothesis? Inside the Viral Saga

Claims that AI had proven the Riemann hypothesis went viral, drawing skepticism. A circulating Alpöge statement revealed Dario Amodei asked him to wait before announcing.

2026-11-20 ~ 2026-11-21 · 2 episodes · 117 posts

Episode 1 · Anthropic Mathematician Claims Claude-Helped Riemann Hypothesis Proof (2026-11-20, 115 posts)

Starting November 20, claims that "AI has solved the Riemann hypothesis" went viral across social platforms, drawing crowds of onlookers, half believing and half doubting. No confirmed primary source exists so far.

Confirmed

  • AI researcher YuchenjUW tweeted tongue-in-cheek that "the Riemann hypothesis has been solved by AI," saying future Fields Medalists will be named Claude, GPT, or Gemini, and mentioned a widely shared claim that a human-readable version of the "proof" runs over 250,000 pages.
  • User yifanzhang posted that "the Riemann hypothesis has been solved" with a link; zephyrz9 reshared it with only an ambiguous "Well..."
  • Science writer Erik Hoel quoted his earlier stunned remarks: someone claimed that with human guidance, Claude did substantial work in the final stages of Riemann-hypothesis-related work. He concluded "we clearly have superhuman math capability," and said this crossed a red line he had drawn—whether AI can produce genuinely novel, creative intellectual output.
  • On Numberphile, Terence Tao once said "with AI's help, the Riemann hypothesis could be solved before 2050"; this clip spread wildly with captions like "it only took 35 days," spawning memes such as a Dario apology statement and "Anthropic casually proving the Riemann hypothesis amid the AI bubble burst."

Unconfirmed

  • Whether the Riemann hypothesis was actually proven, what exactly Claude contributed, and the contents and provenance of the "250,000-page proof"—none are backed by original papers or statements from involved parties. Everything remains rumor and memes.

Why it matters

  • Erik Hoel offered a sober boundary analysis: these proofs "look hard" because finding them with existing tools requires combining vast amounts of technique, and the solution space of math problems is well-defined—two factors that make the task relatively AI-friendly. But concluding "math is easy for AI" from this would be a misjudgment. The Riemann hypothesis is a Millennium Problem; if AI could substantively participate in its proof, it would mark a major leap in AI creativity—which is why the rumor ignited the community so quickly even unverified.

95 more related posts →

Episode 2 · Riemann Hypothesis Reportedly Proven; Announcement Held Back (2026-11-21, 2 posts)

Math circles are abuzz over a rumored proof of the Riemann Hypothesis. Leaked excerpts of Alpöge's statement reveal he asked Dario Amodei late Sunday night whether he could announce it, but was told constraints required him to wait—joked to be tied to Anthropic's impending IPO.