AI solutions introduce new security, privacy, and governance challenges that traditional application security models were not designed to handle. Securing prompts, models, tools, data access, and agent behavior requires a fundamentally different approach.

In this session, you will learn how to secure AI applications and agents built on Microsoft Foundry across their entire lifecycle, from design and development to deployment and runtime operations. We will explore how to establish strong identity and access controls for agents and tools, securely connect AI systems to enterprise data, and apply least-privilege principles using managed identities, role-based access control, and policy enforcement.

You'll also learn how Microsoft Foundry enables enterprise-grade governance at scale, including content safety, grounding controls, auditability, traceability, and posture management. Through real-world architecture diagrams and practical scenarios, you will see how to prevent common AI security failures such as prompt injection, data exfiltration, over-privileged agents, and ungoverned model usage.