Agentic AI and the New Frontier of Identity Attacks

As enterprises strengthen their perimeter defenses, identity has become the crown jewel for attackers—and the emergence of agentic AI is changing the game. Autonomous, goal-driven AI systems can now execute multi-stage identity attacks with unprecedented speed, scale, and adaptability. From harvesting credentials and bypassing MFA to performing autonomous privilege escalation and persistence, agentic AI can operate continuously, learning from every failed attempt and dynamically adjusting tactics to evade detection. This session explores how agentic AI transforms the identity attack surface: orchestrating reconnaissance to map identity hierarchies, exploiting misconfigurations in identity providers, abusing OAuth and service accounts, and coordinating multiple AI agents to maintain access. We’ll dissect real-world-inspired scenarios that showcase AI-driven lateral movement, privilege abuse, and stealthy persistence within identity systems like Active Directory and Entra ID. You’ll leave with a set of practical strategies for defending against agentic AI threats. In the age of agentic AI, protecting identity is not just a security function—it’s the last line of defense.