Nick Clark - The 16 Minute Breach: Why Human-Led Defense Fails Against Agentic AI

As enterprise AI adoption has surged—with a 91% year-over-year increase in transactions—the attack surface has fundamentally shifted. In early 2026, Zscaler ThreatLabz research identified that the median time to critical failure in enterprise AI systems is now just 16 minutes. We have entered the era of "Agentic AI" as a machine-speed attack vector, where autonomous agents handle reconnaissance, exploitation, and lateral movement faster than a human SOC can triage an alert. This session moves beyond the "Shadow AI" conversation to explore the reality of autonomous threats and the "Oversight Gap." We will analyse recent industry-shifting incidents, such as the Bob-p2p finance compromise and ClawHavoc supply chain poisoning, to demonstrate why traditional perimeter-based security is obsolete. Attendees will learn how to transition from human-speed reactive security to AI-powered Zero Trust and proactive AI Red Teaming, ensuring that defence can finally match the velocity of the modern, autonomous attacker.