Securing AI Agents: From Identity Control to Runtime Defense

Speakers

• Pim Jacobs

About this session

As organizations scale automation and agentic AI, AI agents become privileged actors—embedded deep in business critical workflows, data flows, and decision chains. But what happens when an agent is misconfigured, compromised, or starts consuming malicious resources? In this session, we unpack the security architecture of AI agents and expose the often overlooked risks hiding behind autonomous behavior. We walk through how to secure agents end to end using Agent 365’s security stack—from identity and access controls, to policy enforcement, to runtime threat protection. Expect real world examples of common agent misconfigurations, attack paths targeting agent identities and tools, and practical mitigation strategies. You’ll leave with a clear security blueprint to build a trustworthy, resilient agent ecosystem—designed to withstand real world threats, not just theoretical risks.