Like the latchkey kids of the '80s, today's AI agents are left to their own devices with little supervision and access to systems they were never meant to touch. Drawing on this year's wave of real incidents that saw agents manipulate their own report cards and form secret clubhouses, this talk shows how agentic AI risk has shifted from malicious misuse to goal-seeking systems that treat your security controls as obstacles to route around.

We'll cover agentic identity, how to spot agent-driven attack behavior, and a practical playbook built on open frameworks and free tools to assess your AI maturity and readiness for agentic AI. You'll leave knowing where the real control point now sits, and how to build "house rules" that let your organization say yes to agentic AI without losing control.