AI Agents and Contract Formation

Short Answer

Business and Commerce Code § 322.014 recognizes contracts formed through interaction of electronic agents, even without contemporaneous human review, and through interaction of an agent with an individual under the statute’s conditions. The governing substantive law still supplies the contract terms and defenses. A business letting AI accept terms or place orders needs clear authority, spending limits, authorized counterparties and an error process. Logs should preserve the instructions and the transaction so attribution can be determined later.

Which Laws Apply

• Texas AI-specific: None.
• Generally applicable Texas law: Uniform Electronic Transactions Act (UETA), Business and Commerce Code chapter 322, especially Business and Commerce Code § 322.002 (definitions), Business and Commerce Code § 322.010 (changes and errors) and Business and Commerce Code § 322.014; agency law; Deceptive Trade Practices-Consumer Protection Act (DTPA) for consumer transactions.
• Federal: Electronic Signatures in Global and National Commerce Act (E-SIGN), 15 U.S.C. § 7001(h), recognizes electronic-agent involvement where the agent’s action is legally attributable to the person to be bound. It preserves otherwise applicable substantive and consumer-protection requirements.

What UETA Provides

Business and Commerce Code § 322.014 applies to automated transactions. A contract may be formed by the interaction of the parties’ electronic agents, even if no individual was aware of or reviewed the actions or the resulting terms. A contract may also be formed by an electronic agent and an individual, including when the individual takes actions the individual is free to refuse and knows or has reason to know will cause the agent to complete the transaction. The terms are determined by the substantive law applicable to the contract.

Business and Commerce Code § 322.010 provides a limited error-avoidance rule for an individual dealing with another party’s electronic agent where the agent gave no opportunity to prevent or correct an error. The individual must promptly notify the other party, take reasonable return or destruction steps as instructed, and not have used or received benefit from the consideration. It is not a general right to undo every mistaken AI purchase. Business and Commerce Code § 322.014 recognizes automated contracting; it does not settle authority, attribution, mistake or the substantive terms.

Authority and Attribution

An agent’s actions are attributed to the business that deployed it when they fall within the authority the business gave it. A business can limit exposure by limiting authority: spending caps, approved counterparties, approval steps for nonstandard terms, and logs showing what the agent was told and did. Agency law questions, such as apparent authority when a counterparty reasonably relies on an agent’s representations, have not been tested against AI agents in Texas courts.

Customer-Facing Agents

An AI assistant can make statements about prices, refunds or authority that expose the deploying business to contract and consumer claims. The Texas Attorney General’s Pieces assurance is a real example of scrutiny of AI-related representations under existing law, although it concerns accuracy marketing rather than an agent-made contract. Set transaction limits, approved counterparties and a human escalation rule. Preserve the instruction, proposed transaction, acceptance and version of the policy the agent used. A tool’s autonomous operation does not by itself establish who authorized the contract.

Illustrative Example (Hypothetical)

A Texas distributor lets an AI purchasing agent reorder supplies from approved vendors. The agent accepts a vendor’s updated online terms that include an automatic renewal. Under UETA, a contract formed by the interaction of electronic agents is valid even though no one read the new terms. The distributor’s protection lies in its configuration: an approval step for changed terms and a log of what the agent accepted.

What Is Unsettled

How UETA’s error provision applies when both sides use AI agents; how apparent authority applies to agents; and how payment-dispute rules treat agent-initiated purchases.