The AI Was Granted Access – No One Saw What It Did Next [Intermediate]

AI agents today operate autonomously across production systems, cloud environments, and databases – often with far-reaching permissions that no one monitors in real time. Classic access controls check who gets in, but not what actually happens within a session. That's exactly where the new attack surface emerges. This talk shows how effective zero trust security works today for humans, non-human identities, and agentic AI alike – not as a vision for the future, but in productive use, already today. You'll learn how organisations are moving from standing privilege to just-in-time access, making every single action traceable, and maintaining control without sacrificing speed or productivity.