What happened
Anthropic announced (Aug 21, 2026) that Claude Security scans now run on Claude Mythos 5, its most cyber-capable model, in public beta for all Claude Enterprise customers — scanning GitHub repos for vulnerabilities with CWE classification, severity/confidence ratings, and suggested patches (human-reviewed only). Anthropic is also integrating Mythos 5 into partner security products, launching a $35M Defender Advantage Fund for OSS security, and expanding its Cyber Verification Program.
Why it matters
This is the first time Anthropic's most capable frontier cyber model is made available to enterprise defenders broadly (via controlled output, not direct model access), materially raising the bar for AI-assisted vulnerability discovery/remediation while managing dual-use risk.
Applicability
Enterprise AppSec/security teams already on Claude Enterprise should evaluate the beta now; vendors should watch for Mythos 5 partner-integration APIs.