Anthropic's Project Glasswing Uncovers 10,000 High-Severity Vulnerabilities in Critical Software

Anthropic’s Project Glasswing — a defensive AI initiative using Claude Mythos to audit globally critical software — has surfaced more than 10,000 high- or critical-severity vulnerabilities since launching last month. The effort targets what Anthropic describes as “systemically” important software and operates at a scale that manual review cannot match.

This expands significantly on the earlier Anthropic-Mozilla Firefox collaboration (271 vulnerabilities). For security teams: AI-assisted auditing at this throughput is beginning to reshape coordinated disclosure pipelines; expect Glasswing findings to appear across major open-source projects and to accelerate patch cadences for critical infrastructure.