Trending: Anthropic claims AI generated proof of the Riemann hypothesis Hottest news
Employee says Claude chatbot produced a 250,000‑page proof of the famed prime‑number problem shortly after the company's IPO.
Image: BBC News Download
Levent Alpöge, a mathematician on Anthropic’s staff, told reporters on Friday that Claude, the firm’s chatbot, churned out a full‑blown proof of the Riemann hypothesis. He made the claim just one day after Anthropic’s stock debuted on the NYSE, a headline‑grabbing moment for the AI start‑up.
Anthropic’s claim and the alleged proof
BBC News relayed that Alpöge logged about six weeks tinkering with a private Claude instance, coaxing it into a proof that now stretches beyond 250,000 pages of Lean code, roughly the size of twelve thick mathematics tomes. He called the breakthrough “a dawn across science of something far greater.” The hypothesis, first sketched by Bernhard Riemann 170 years ago, predicts a precise rhythm in how prime numbers line up and remains the crown jewel of unsolved math problems.
Alpöge said he rang Anthropic CEO Dario Amodei on Sunday to hash out the timing. “The precise timing is horrible,” he complained, pointing to the fresh IPO. “I am clearly unilaterally responsible for feeling I just could not live if I waited even longer. There is an obvious cynical view, to do with the timing, but it is false.”
Academic reaction and verification challenges
Fields medalist Terence Tao warned that digesting a manuscript of that magnitude will take “considerable time.” He added, “I imagine the digestion process will take some time, though in this case at least we have a human author to communicate with to assist with this task.” The sheer length and its reliance on automated proof assistants mean peer review could stretch for months.
A coalition of mathematics scholars wrote a joint letter urging restraint. They argued that flashing AI breakthroughs before rigorous validation threatens the discipline, calling the trend “detrimental to the science of mathematics” and warning of a “general threat to intellectual work” when AI output diverges from traditional research goals.
Broader AI activity in mathematics
Anthropic’s announcement sits on a growing pile of AI‑driven attempts to crack old problems. In September, OpenAI disclosed that an internal team using a private tool solved a piece of the Navier‑Stokes equations in 88 hours, marshaling more than 10,000 autonomous agents. The company later claimed progress on 372 additional high‑level challenges, though those findings are still under expert scrutiny.
Both firms tout wins in physics and other sciences, but scholars keep stressing that independent verification is the only path to accepting any AI‑generated solution as definitive.
What happens next
- Mathematicians will launch a systematic review of the Lean code to see whether the Riemann hypothesis truly folds under Claude’s pen.
- Anthropic plans to publish the full proof, giving outsiders a chance to rerun the calculations and test every step.
- The verdict will shape how AI tools sit inside mathematical research and could steer future funding, policy and regulation of AI‑assisted discovery.
Further reading
For more context on this story, explore these trusted outlets.
