Our prior article argued that agentic commerce is exposing legal fault lines that ordinary e-commerce could often paper over. This article takes up the first of them: identity and authentication.
That issue sounds narrower than it is. In agentic commerce, the first serious legal problem is not assent, apparent authority, or loss allocation. It is more basic: When an AI agent enters a transaction flow, who exactly is the merchant dealing with?
In ordinary e-commerce, the answer is usually functional rather than philosophical. A merchant sees a logged-in account, a recognized device, a payment credential, and a user completing a purchase through a familiar interface. Identity, authority, and action may not be perfectly aligned, but they are aligned closely enough that the system can treat them as though they come from a single commercial actor.
