Mathematicians say Anthropic's Claude-made Riemann hypothesis proof compiles in Lean
Independent researchers, including MIT's Andrew Sutherland, compiled the Lean code without issues. A final check called a comparator run is still pending.
What happened
An Anthropic mathematician, Levent Alpöge, says he used the company's Claude AI model to prove the Riemann hypothesis. It is a problem from 1859 about a hidden pattern in prime numbers. Other mathematicians have now checked the computer version of the proof and found no errors so far.
Why it matters to you
If it holds up, an AI system would have solved one of the most important open problems in mathematics.
The takeaway
The proof looks sound, but one last safety check is still unfinished.
What it means for you
If you build software
Lean compilation is the main check here, and the comparator run tests for redefined symbols and kernel bugs.