An AI agent orders the wrong goods, accepts unfavourable terms or agrees a price no human would. Is there a contract?
The starting point
An AI system is not a legal person. It cannot be an agent in the legal sense, because agency is a relationship between persons. The question is instead whether the person or business that deployed it is bound by what it did.
English law
Courts have accepted contracts concluded by automated systems. In Quoine v B2C2, a 2020 Singapore appeal decision applying common law principles, trades executed by algorithms were binding, and the question of mistake was judged by the knowledge and intentions of the programmers when they wrote the code. The Law Commission has said existing principles can generally deal with automated contracting. A business that lets an agent negotiate on its behalf is likely to be held to the outcome, unless the other side knew of an obvious error.
United States
The Uniform Electronic Transactions Act, adopted in almost every state, provides that a contract may be formed by the interaction of electronic agents, even if no human reviewed the terms. The federal E-SIGN Act takes the same approach.
European Union
Contract formation is national law, but EU e-commerce rules require that contracts can be concluded electronically. Consumer law adds duties such as clear information and, for most distance contracts, a right of withdrawal.
If you give an AI agent a credit card, assume you have also given it your signature.
What to do
Set hard limits on what an agent can commit to: value caps, approved counterparties, and human approval above a threshold. Keep logs of what it did and why. Check that platform terms allow agent use.
This article is general information on the law as at November 2026. It is not legal advice and has not been updated since that date. See the terms of use.