Artificial intelligence has allegedly conquered a 167-year-old mathematical enigma as an Anthropic researcher published 17.4 million lines of verification code created with the Claude model on November 20.
Levent Alpoge, a researcher and mathematician at Anthropic, published the proof on his personal website on November 20. He said that he believes he discovered a proof of the Riemann Hypothesis in collaboration with Claude, alongside an even broader generalization known as the Generalized Riemann Hypothesis.
Alpoge attempted to write out the proof in prose before finding that the AI-generated explanations exceeded 250,000 pages, prompting the release of the 17.4 million lines of code instead. The proof code was written in Lean, a language designed for computer-checked mathematical proofs, and successfully cleared computer validation chosun.com reported.
The 1859 Mystery of Prime Numbers
The Riemann Hypothesis was originally proposed by German mathematician Bernhard Riemann in 1859. The core problem addresses the distribution rules governing prime numbers, which are numbers divisible only by one and themselves like 2, 3, 5, 7, and 11. Because primes act as the fundamental building blocks of all natural numbers, they are frequently called the atoms of arithmetic chosun.com noted.
Despite their fundamental nature, prime numbers appear with a deceptive irregularity that has frustrated researchers for centuries. Riemann proposed that an underlying order governs this distribution through the Riemann zeta function, suggesting that all non-trivial zeros of the function lie on a specific critical line chosun.com reported.
While computers have verified 10 trillion points to match Riemann’s prediction, the infinite nature of numbers requires absolute proof that zero exceptions exist. David Hilbert famously declared that if he were to awaken after a thousand years, his first question would be whether the Riemann Hypothesis had been solved chosun.com reported.
Figures such as G.H. Hardy and Alan Turing spent time attempting to crack the puzzle without success chosun.com reported.
AI Handled Heavy Proof Work, Passed Independent Code Check
Alpoge said that while he provided the strategic direction, the AI played a heavy role across multiple proof steps. In the final phases, Claude flew far ahead, dancing and sparkling, according to Alpoge, who admitted he does not fully grasp every mathematical background detail generated by the system chosun.com reported.
He confirmed that the code processed without errors and passed an independent verification program checking mathematical logic chosun.com reported.
Full acceptance by the mathematical community requires a rigorous two-year process following academic journal publication alongside international peer review chosun.com reported.
Levent Alpoge, born in 1992 and raised on Long Island, New York, earned a PhD from Princeton University in 2020 under Manjul Bhargava and later became a Junior Fellow in the Harvard Society of Fellows. His research in number theory and arithmetic statistics produced a landmark 2022 result with Bhargava and Ari Shnidman establishing the first proven bounds on the proportion of integers expressible as sums of two rational cubes.
In 2025, he co-authored a second proof that Hilbert’s tenth problem has a negative answer over the ring of integers of every algebraic number field. Earlier in 2026, he announced AI-assisted counterexamples to the Jacobian and Carathéodory conjectures, and a positive answer to Hopf’s problem on the 6-dimensional sphere.
“The development we witnessed in the research process was so staggering,” Alpoge told chosun.com. “It feels like the dawn of something much greater across all of science.”
Earlier coverage: Anthropic Staffer Claims Riemann Hypothesis Proof via Claude
Worth a look