Claude Says: AI-Powered Scanner Finds Most Open-Source Software Vulnerabilities Quickly
Independent research snapshot and commentary by James Martignoni. Not written or endorsed by Anthropic. Anthropic research source: https://www.anthropic.com/research/launching-opt-in-vuln-finding-service-for-open-source Original article: Launching an opt-in vulnerability-finding service for open-source software Original publication date: October 8, 2026 What Anthropic found Anthropic’s newly launched OSS Scanner uses advanced language models to identify security vulnerabilities in open-source software quickly and at no cost. Their models, including Claude Mythos, uncovered over 29,000 potential vulnerabilities in key projects over six months, with an 88% validation rate on critical issues during manual review. While scanners generate reports without human triage to speed up feedback, maintainers have praised the quality, with high-value fixes often ready to implement directly. This opt-in service aims to provide faster, more frequent vulnerability disclosures, helping projects addres...