Anthropic Employee Claims AI-Assisted Solution to Millennium Prize Problem
Levent Alpöge, a technical staff member at Anthropic, announced on November 20, 2026, that he solved the 167-year-old Riemann Hypothesis using the company's AI model.
70-100 BALANCED · 40-69 MIXED · 0-39 LEANING.
Euronews
Levent Alpöge, a technical staff member at Anthropic, announced on November 20, 2026, that he solved the 167-year-old Riemann Hypothesis using the company's AI model. Alpöge released 17 million lines of Lean code to support his claim. This work also covers the generalized Riemann Hypothesis. Two proof-checking systems accepted the code. Clay Mathematics Institute still lists the problem as unsolved. Independent human experts must verify these AI-generated arguments. This rigorous step precedes the awarding of the $1 million prize.
Context
The Riemann Hypothesis concerns the distribution of prime numbers. It has remained a significant challenge in mathematics for over 160 years. Efforts to prove this hypothesis drove the development of new mathematical techniques. Artificial intelligence models now assist with complex mathematical problems. Computational tools help address theoretical obstacles that resisted traditional human-led inquiry.
The Riemann Hypothesis was first formulated in 1859. The Clay Mathematics Institute designated the hypothesis as one of seven Millennium Prize Problems in 2000. The institute offers a $1 million reward for a valid solution. Technological advancements now apply to other complex mathematical challenges. OpenAI reported in September 2026 that its technology assisted in a breakthrough regarding the Navier-Stokes equation.
Supportive perspectives
The Riemann Hypothesis concerns prime number distribution. Bernhard Riemann proposed it in 1859. The Clay Mathematics Institute named it one of seven Millennium Prize Problems in 2000. The institute offers a $1 million reward for a proof.
Levent Alpöge released 17 million lines of Lean code on November 20, 2026, to document his proof of the Riemann Hypothesis. He stated on November 20, 2026, that he believes he found a valid proof using Claude, an AI model from Anthropic. MIT mathematician Andrew Sutherland confirmed that the published code passed several verification checks.
Rutgers University professor Alex Kontorovich noted that the computer's verification process is 'extraordinarily picky.' He stated that if code meets these rigorous standards, there is a 'very, very good chance' it will also satisfy his requirements.
Levent Alpöge also claimed to have proved the generalized Riemann hypothesis.
Anthropic held an initial public offering on November 19, 2026, raising $81 billion at a $1.5 trillion valuation. The company held an IPO closing bell ceremony at the Nasdaq MarketSite in New York on the same day.
Critical perspectives
As of November 21, 2026, the Clay Mathematics Institute continues to list the Riemann Hypothesis as an unsolved problem. Martin Bridson, president of the institute, stated that the journey from the announcement to formal verification will be a long and difficult process.
Levent Alpöge did not produce a traditional academic paper to accompany his work. Alpöge acknowledged that he does not yet fully understand every argument the Claude AI model generated to produce the result. The proof currently lacks a complete human-readable logic.
Alex Kontorovich, a mathematics professor at Rutgers University, stated that while the computer verified the code, human understanding of the proof is still necessary. He noted, as NPR reported, that he does not yet understand the proof himself.
Sources verified by Neutral
This story is built from the reporting below. Tap a source to read the original.
Also part of
More in The Science Horizon
See all 44 stories in this arc ›Get the full picture on the Neutral News app
Both sides of every story on one page, with a neutrality score on top. Follow stories as they unfold. On Android and iPhone.
Already installed? Open in the app