Independent verification confirms AI-assisted proof of the Riemann hypothesis

4 min read
Source: USA Today
Independent verification confirms AI-assisted proof of the Riemann hypothesis
Photo: USA Today
TL;DR

Levent Alpöge of Anthropic announced on Nov. 20 that he used the Claude AI model to prove the Riemann hypothesis, a 167-year-old problem concerning prime number distribution. Independent mathematical communities confirmed the proof’s validity on Nov. 21, though debates persist regarding AI’s role in research and the implications for human mathematicians.

Key points

  • Levent Alpöge, a mathematician at Anthropic, announced on Nov. 20 that he collaborated with the Claude AI model for approximately six weeks to develop a proof of the Riemann hypothesis.
  • The proof was submitted as computer code in the Lean programming language, which allows for automated verification of mathematical logic.
  • On Nov. 21, the Lean Focused Research Organization and the Mathlib Initiative independently verified the code, stating they had 'full confidence' in the correctness of the proof.
  • The Riemann hypothesis, proposed by Bernhard Riemann in 1859, is one of seven Millennium Prize Problems, each carrying a $1 million prize from the Clay Mathematics Institute.
  • Alpöge also claimed to have established the Generalized Riemann Hypothesis, extending the original problem to a broader family of mathematical functions.
  • The announcement follows a period of intense AI activity in mathematics, including OpenAI’s release of over 700 AI-generated manuscripts in October 2026.

Background

This development follows a series of AI-driven mathematical breakthroughs in 2026. In October, OpenAI released over 700 AI-generated manuscripts, claiming to solve 372 open problems, though some were later retracted due to errors. This release included progress on a 'quasi-Riemann hypothesis,' a looser version of the original problem. The current proof represents a significant escalation in AI capability, moving from approximate solutions to a definitive proof of one of the most famous unsolved problems in mathematics.

How outlets are covering it

USA Today and The New York Times highlight the collaborative nature of the proof, noting Alpöge’s statement that Claude 'far outpaced' him near the finish line. Scientific American emphasizes the rigorous independent verification process, citing mathematicians Andrew Sutherland and Kevin Buzzard, who confirmed the Lean code compiled without errors and did not exploit system bugs. The Guardian offers a contrasting view through mathematician Ken Ono, who expresses concern about the future of human mathematicians in an AI-dominated field. Ono notes that while AI can formalize and verify proofs, human judgment remains essential for understanding the underlying ideas. He also highlights the ethical concerns raised by other mathematicians, such as Brian Conrad, who criticized Alpöge for releasing a proof he did not fully understand. Additionally, The New York Times reports on a competing, incomplete proof by Alessandro Masullo, which used a different approach and did not involve AI, highlighting the diversity of methods in the field.

Why it matters

The verification of the Riemann hypothesis proof marks a pivotal moment in the intersection of artificial intelligence and mathematics. It suggests that AI may be capable of solving problems that have eluded human mathematicians for centuries, potentially reshaping the field. The proof could also have broader implications for other areas of mathematics, as noted by mathematician Hector Pasten, who compared it to a 'Rosetta Stone' for connecting different areas of math. However, the event also raises significant questions about the role of AI in research, the credit given to human contributors, and the future of mathematical education and career paths.

What to watch

The Clay Mathematics Institute continues to list the Riemann hypothesis as unsolved, indicating that the proof has not yet been officially recognized for the $1 million prize. Mathematicians are still assessing the broader implications of the proof and the methods used to develop it. The debate over AI’s role in mathematics is likely to continue, with institutions and funding agencies needing to decide how to support and reward research in this new era. Additionally, the competing proof by Alessandro Masullo may undergo further review, potentially offering alternative insights into the problem.

Share this article

Want the full story? Read the original reporting

Read on USA Today