AI agents are quickly becoming the new 'virtual employees' of enterprises... and attackers know it. In this interactive attacker vs defender session, we follow a real world attack path to show how a malicious actor manipulates an enterprise agent and how each security layer responds in sequence. You’ll see how the AI Gateway acts as the security checkpoint, inspecting and governing agent traffic. How Microsoft Foundry enforces runtime guardrails, safe tool use, and policy based behavior controls inside the agent’s execution environment. How Purview protects the “vault” of sensitive data, and how Defender for AI is used for threat detection, alert triggering and incident response capabilities. This session demonstrates how to stop attacks before they become incidents using proactive measures, investigation flows and observability traces. This will give you a clear view of how to architect AI agents with robust, production grade, end to end security.