Claude Opus 4.6 Launches Enhanced Security Capabilities to Validate 500+ Critical Vulnerabilities

Anthropic's Claude Opus 4.6 is presented as a significant advance in automated vulnerability discovery, using semantic code analysis and commit-history reasoning to find hundreds of high-severity bugs in open-source projects (notably GhostScript, OpenSC, and CGIF). The report contrasts this approach with traditional fuzzing, illustrates specific findings (incomplete patches, strcat-related buffer overflows, and LZW-related heap overflows), and warns of dual-use risks while noting built-in probes to mitigate misuse and the need for faster patch cycles.