Agentic AI has fundamentally altered the threat landscape. Today's adversaries deploy autonomous agents that bypass traditional bot detection through human-like browsing patterns, natural language generation, and adaptive behavior that learns from each blocked attempt. These same capabilities power legitimate business automation, customer service operations, and productivity tools your users depend on. The ratio of non-human to human identities is already around 80:1 and is expected to grow to many times over in the coming months as Agentic AI becomes mainstream. Enterprises need to figure out how to manage good vs. malicious agents while maximizing benefits from AI related technologies. 

This session examines the technical and business realities of this paradox through three critical lenses:

The Threat Evolution: We'll dissect real-world attack patterns we're observing at scale from LLM-powered phishing campaigns that generate contextually perfect spear-phishing at industrial scale, to credential stuffing operations using AI agents that solve CAPTCHAs, navigate MFA flows, and mimic human session behavior. You'll see how attackers weaponize tool-calling frameworks, chain multiple AI services, and use reinforcement learning to optimize attack efficiency in real-time.

The Detection Challenge: Traditional signals fail against agentic threats. Mouse movements can be synthesized. Behavioral biometrics can be learned. Session patterns can be replicated. We'll examine why rule-based systems, simple ML models, and legacy bot detection are inadequate, and explore the technical architecture required for real-time intent analysis including behavioral fingerprinting that distinguishes scripted automation from human-directed agents from fully autonomous AI actors.

Strategic Defense Principles: We'll outline the architectural thinking required to defend against agentic threats, moving from binary allow/block decisions to continuous trust evaluation, from static rules to adaptive risk models, and from point-solution detection to platform-wide orchestration. You'll understand the key components needed in a modern defense stack and how they must work together to make intelligent decisions about agent legitimacy without breaking legitimate use cases.