Glasswing Secured the Code. The Rest of Your Stack Is Still on You

The article argues that Anthropic's Project Glasswing and its Mythos Preview model demonstrate a new class of automated vulnerability discovery that 'reads' code like a human and found long-lived flaws (e.g., an FFmpeg bug) missed by millions of fuzzer runs; it warns that while code vulnerabilities matter, the larger, persistent organizational risk remains misconfigurations, stale credentials, and incomplete posture visibility, and urges teams to validate inventories and readiness ahead of coordinated disclosures.