How Mythos changes the AppSec calculus

Anthropic's unreleased 'Mythos' model is described as capable of autonomously finding and chaining software vulnerabilities, potentially shrinking the vulnerability-to-exploit timeline from months to minutes; the article warns this will democratize advanced attack capabilities but also notes limitations (compute/access, false positives, limited scope) and recommends adopting a layered AppSec strategy—discovery, analysis, remediation, runtime, and context—rather than buying a single AI product.