Your riskiest employee might not be human

AI agents don't clock in, but they authenticate, access data, and act at machine speed on someone's behalf - and most security programs still aren't built to see them. This session reframes agentic AI as an insider risk problem, not a new category to bolt onto an already-crowded stack. Expect real patterns from the field, a simple framework for knowing what's running, who deployed it, and what it can reach - and an answer on what governing this actually looks like in practice.