Zero trust applied to AI agents requires identity to be established, but the remaining privileges such as least privileged access and assume breach are vague. This session will dive into what it practically means to apply zero trust to AI agents and what controls are most effective.