Overview
- Levent Alpöge, a mathematician at Anthropic, published a four‑page note and a Lean formalization and said he developed the proof working with Anthropic’s language model Claude.
- Alpöge also claims his approach proves a stronger generalized form of the Riemann hypothesis that applies to a family of related functions.
- The work is public but unverified, and mathematicians must check the informal reasoning and confirm the Lean code correctly encodes every logical step before accepting the claim.
- A separate long draft by another researcher was posted shortly after Alpöge’s announcement, so multiple independent lines of unverified proofs are now circulating in the community.
- The Riemann hypothesis concerns the nontrivial zeros of the zeta function and the distribution of prime numbers, and a confirmed proof would resolve one of the Clay Institute’s Millennium Prize Problems and change how researchers view AI’s role in deep mathematical discovery.