Delegating to Machines: The Legal Fine Print

This session is part of the Think Tank series. Think Tank sessions are 45-minute breakouts with a 15-minute Q&A where experts in law and consultancy share their expertise on relevant and thought-provoking topics.

When an AI agent clicks “I agree” or completes a purchase, centuries-old agency doctrines collide with autonomous systems. When an AI agent acts, what determines its authority? How can platforms strengthen apparent authority where they want it (purchases, terms, consents) and limit it where they don't (misrepresentations, dark patterns, unauthorized account changes)? This panel brings together AI agent developers and the platforms they access to examine AI agent apparent authority and allocation of liability. The panel will explore how contracts, user experience, and technical controls shape responsibility, and how human-in-the-loop safeguards bolster agent authority.

What you will learn:

• Agency law is being stress-tested by AI. When an AI agent clicks “I agree” or makes a purchase, courts will have to decide whose authority that action represents — user, platform, developer, or no one. Companies must define agency roles now, or courts will do it for them.

• Human-in-the-loop is legal architecture. Confirmation steps and authentication gates are not just security features — they build a defensible story of reasonable reliance and apparent authority. Where you place (or omit) those gates shapes liability.

• Early design choices will shape the law. Through contracts, user interface flows, and technical controls, companies can influence how courts apply centuries-old agency doctrine to agentic AI. The governance frameworks built today will determine exposure tomorrow.

Sponsored by: Cooley