Project Glasswing Proved AI Can Find the Bugs. Who's Going to Fix Them?

**The article warns that Anthropic's Project Glasswing / Mythos demonstrates AI-class vulnerability discovery and exploit chaining at scale—finding long-lived bugs, constructing multi-bug exploit chains, and achieving high success rates—while remediation processes are far too slow; it argues defenders must adopt autonomous, environment-specific validation (such as Picus Swarm) to triage and validate which findings are truly exploitable before attackers weaponize them.**