Your firewall can stop a hacker, but can it stop your own AI agent? As enterprises deploy autonomous agents, they face a new threat: "Authorized Rogues" i.e. valid agents tricked into executing malicious workflows. In this session, we move beyond high-level policy to share battle-tested architectural patterns for securing Agentic systems. You will learn why traditional IAM fails against probabilistic reasoning and how to implement "Governance Circuit Breakers", the deterministic controls that halt cascading failures in real-time. Join this session to transform your AI governance from a passive policy into an active, architectural defense.