Anthropic launched OSS Scanner on October 8, offering free AI-driven security scans for critical open source projects using Claude models. Core maintainers apply via GitHub PRs. Over six months, the system identified 29,000 candidate vulnerabilities, with 6,000 manually reviewed. Validation of 97 high-severity findings across 48 projects showed an 88% accuracy rate for CVD inclusion. The service uses coordinated disclosure and a fast-track option, continuously refining detection as models evolve.