Everyone is calling AI a threat and selling "AI Security" as the answer. This keynote argues both are wrong.

AI isn't a new category to bolt onto your stack. It's working its way into every layer you already own, and identity is where that gets dangerous first. Agents inherit their creator's permissions and move at machine speed, which turns the standing access we've quietly tolerated for years into a live, exploitable risk.

This session provides a framework for controlling access in the agentic era, and shows why the instinct to buy a new category will leave security leaders more exposed, not less. We'll cover the practical tools and methodologies that make it work, including Just-in-Time access, Zero Standing Privileges, ephemeral access, and the role of context and audit in governing actors that move faster than any human can review. If you own identity, govern agents, or are being told to "secure AI" without a clear definition of what that even means, come find out where this is really headed.