When Agents Direct Agents: Who Do You Actually Regulate?

Financial regulation has always assumed that decisions can be traced through a clear chain of human accountability. But as AI agents begin directing other agents, that assumption starts to break down. Agentic AI won't just automate tasks—it has the potential to reshape how financial institutions are organised. Decision-making may shift between people and AI systems, reporting lines become more fluid, and the traditional structures underpinning governance, accountability and supervision begin to evolve. In that world, who is ultimately responsible for a decision? What exactly is a regulator supervising?

Bringing together senior policymakers and regulators, this session explores how supervisory frameworks—from senior manager accountability regimes to board oversight and the three lines of defence—need to adapt for organisations where humans increasingly govern systems that, in turn, govern other systems. The discussion will examine how institutions can demonstrate accountability, maintain effective oversight, and provide regulators with meaningful traceability in increasingly agent-driven operating models. Rather than focusing on individual AI decisions, the panel asks a broader question: how should governance evolve when the organisation itself becomes dynamic?