Agentic AI reached production environments in just six months. That’s a fraction of the two years it took for traditional generative AI adoption. This acceleration has created security challenges that require new approaches to protection and monitoring.
At RSAC 2026 Conference, we spoke to Jimmy White, VP of AI at F5 Networks, about why conventional security testing methods fail with agentic AI systems and what organizations must do differently to protect their production environments. We discuss critical insights about the fundamental shift from AI that only “thinks” to agents that can take real-world actions through tools and APIs.
We discuss why runtime security has become non-negotiable, how APIs represent the most vulnerable attack surface in agentic systems, and what practical steps your organization should take today. White shares perspectives on performance trade-offs, explains deployment options ranging from cloud services to air-gapped installations. We also get into why traditional red teaming approaches simply don’t work with non-deterministic AI behavior.
Key insights:
- The dramatic timeline compression from generative AI to agentic systems and what it means for security teams
- Why monitoring agents beats traditional testing approaches for non-deterministic AI
- How AI coding assistants amplify both productivity and security risks exponentially
- The evolution beyond Model Context Protocol and direct API access challenges
- Performance considerations that change the security inspection equation
- Deployment strategies for highly regulated industries requiring complete control
- The surprising analogy between monitoring humans and monitoring AI agents
- Practical first steps for organizations at any stage of AI adoption
Whether you’re just beginning to explore agentic AI or already running systems in production, this interview provides interesting insights into building security postures that can evolve alongside rapidly advancing AI capabilities. Learn why the second-best time to implement AI security is right now, and what specific actions will provide the foundation for protecting your organization’s AI-powered future.