Claude Mythos Expansion Planned Through Claude Code, Security Tools

Anthropic is preparing a staged commercial rollout of its powerful Claude Mythos (Mythos 1) model after an accidental CMS leak exposed about 3,000 internal assets; the company ran Project Glasswing with partners to apply Mythos-class models defensively, reportedly identifying over 10,000 high- or critical-severity vulnerabilities across widely used open-source software and publishing coordinated disclosures (1,596 vulnerabilities across 281 projects with 97 patched). Anthropic is positioning Mythos for gated integration into Claude Code and Claude Security while iterating safety controls and a redesigned vulnerability dashboard for enterprise users.