Anthropic’s Claude Mythos Preview has demonstrated autonomous discovery of vulnerabilities across major operating systems and browsers, including flaws that persisted undetected despite extensive automated testing. A coalition of 12 major technology vendors, Project Glasswing, is deploying Mythos defensively to close these gaps before adversaries develop equivalent capability. The strategic signal is structural: AI-assisted vulnerability discovery has outpaced the remediation pipelines of most enterprise security programs, creating an expanding window of exposure that threat actors, including state-sponsored groups, are working to exploit.