How AI is forcing cybersecurity to become autonomous: Armadin's approach

In a video interview with a16z, Kevin Mandia—who spent 30 years in security and founded Mandiant—explains why he started Armadin to address a fundamental shift in cybersecurity. AI enables attackers to probe thousands of exploit paths simultaneously, making human-speed detection and response obsolete. Traditional penetration testing can no longer keep up with machine-driven threats.

Armadin's model uses AI to continuously attack customer systems to find zero-day vulnerabilities before adversaries do. Mandia argues that defense must become autonomous to match attack speed, and that the entire security operations center (SOC) architecture will need to evolve. He shares lessons from discovering over 90 zero-days in production environments this year and discusses why humans cannot remain in the detect-and-respond loop as AI accelerates both attack and defense cycles.