Mathematics' most famous unsolved problem may finally be dead, and it wasn't a human who delivered the final blow. Levent Alpöge spent six weeks collaborating with Anthropic’s non-public Claude model to crack the Riemann hypothesis. This 170-year-old conjecture about prime number distribution has stumped geniuses for generations, yet an AI now holds the key. The resulting textual proof is so vast that it spans more than 250,000 pages when written in Lean computer language.

Anthropic's Claude solves Riemann Hypothesis

Alpöge didn’t just run a script; he coordinated directly with Anthropic CEO Dario Amodei on Sunday to time this revelation perfectly. The strategy was deliberate: aligning the announcement with the company’s upcoming IPO to maximize impact. While OpenAI previously claimed victories using thousands of agents for fluid dynamics, this feat relies on a single advanced model. The scale of the output challenges our understanding of what constitutes authorship in complex logical structures.

Anthropic news photo

Despite the breakthrough, verification remains a monumental task for the scientific community. Pioneering mathematician Terrance Tao noted that checking such a massive document will take significant time. However, having a human author like Alpöge provides crucial context that aids the review process. This hybrid approach bridges the gap between raw computational power and human intuition.

Anthropic AI

We are witnessing the moment where code begins to outpace centuries of manual calculation. If humans can no longer verify every line of logic, how do we trust truth? The era of solitary genius might be ending, replaced by collaborative intelligence.