Faye Murray

Most AI governance frameworks were built for systems that recommend. Agentic AI introduces systems that can plan, act, use tools, and adapt autonomously, exposing gaps in governance models designed around discrete decisions, clear human authorship, and errors that can be caught in time. This session explores the four assumptions traditional AI governance relies on, why agentic systems break them, and what organisations need to build instead. It will cover decision boundaries, meaningful human accountability, reversibility of actions, and the added complexity of multi-agent systems. Attendees will leave with a practical framework for moving from model governance to action governance, matching controls to the real risks of autonomous AI in production.