AI Meets Cryptography 1: What AI Found in Cloudflare's Circl

zkSecurity's AI audit agent discovered seven real bugs in Cloudflare's CIRCL cryptography library, including critical issues like float64 precision loss in threshold RSA and access-control breaks in attribute-based encryption, all of which have been fixed upstream. The findings come from experiments using large language models with expert-maintained skills to scan open-source cryptography projects, with human validation confirming exploitability before disclosure. The research demonstrates that AI tools can effectively identify severe cryptographic vulnerabilities while revealing insights into how LLMs reason about security, though severity assessments by AI often differ significantly from expert evaluation.