The announcement, which centers on one of the most significant unsolved problems in mathematics for 167 years, has sparked widespread debate regarding the intersection of artificial intelligence and formal mathematical verification.
These zeros determine how far the actual distribution of prime numbers can deviate from the average. Rather than a traditional mathematical paper, the submission consists of formal code written in Lean, a computer-based proof system designed to verify mathematical reasoning.
Collaboration and Technical Complexity
Alpöge characterized the project as a collaborative effort. According to his account, he approached the problem based on an idea from his own field before utilizing Claude to sequence a series of mathematical operations. The researcher utilized a non-public version of Claude that he credits with significant contributions, noting that in the final phases, the model’s output surpassed his own personal contributions.
The scale of the generated proof is substantial. According to a footnote in the document, the Lean formalization spans more than 250,000 pages of computer code. Alpöge acknowledged that the current document is raw and that some descriptions are paraphrases from Claude that may require adjustment, according to Unite.
He admitted that he does not yet fully understand all the mathematical concepts generated by the AI, a transparency that New York University mathematics professor Nestor Guillen praised, arguing that this approach positions AI as a supportive tool rather than a replacement for human intellect.

Verification and Millennium Prize Status
However, the path to claiming such a reward is rigorous; the institute requires publication in a recognized scientific journal and broad acceptance by the mathematical community over a period of at least two years. The president of the Clay Mathematics Institute described the announcement as being of historical importance but cautioned that there is a “long way to go” before the proof is fully understood and verified.
Independent validation is currently in its early stages. Alpöge himself has emphasized that his work is an opening point rather than a final destination, and he expects that simpler methods for arriving at the result will eventually emerge.
Broader Context and Competitive Environment
The timing of the claim has drawn attention, occurring shortly after a major financial milestone for Anthropic. This development follows a similar announcement in September where OpenAI claimed to have used AI to generate a solution for the Navier-Stokes problem, another Millennium Prize challenge.
That event faced controversy, with some mathematicians questioning the methodology and the involvement of researchers like Alpöge and Tristan Buckmaster, who had previously worked on related problems.

The Clay Mathematics Institute, a private foundation established in 1998 by Landon T. Clay, manages these high-stakes challenges as part of its broader mission to disseminate mathematical knowledge. While the organization supports research through fellowships, conferences, and workshops, it is most widely recognized for the Millennium Prize Problems, which it established on May 24, 2000.
These seven problems were selected as significant classic questions that had resisted solution for years, with the institute offering a US$1,000,000 award for the first proven solution to each. In creating these prizes, the institute drew a historical parallel to the famous problems proposed by David Hilbert in 1900.
The broader scientific community remains split on the role of AI in these breakthroughs. While some see it as a necessary evolution in research, a group of mathematicians recently warned that tech companies may be using these claims to demonstrate the sophistication of their products at the potential expense of mathematical integrity.
Anthropic has previously demonstrated Claude’s progress in this field; on August 10, 2026, the company reported that an unreleased version of the model had improved the known lower bound for the proportion of zeta-function zeros satisfying the Riemann hypothesis from 41.6% to 67.2%. Alpöge’s current claim suggests that he has now extended this work to cover both the standard Riemann hypothesis and a generalized version of the problem.