About Agent Authority
A legal reference on what happens when AI agents act for companies.
What it covers
When an AI agent acts for a company, a few questions decide what happens next. Was the agent allowed to do it? Did someone with authority decide to let it? What was promised about what it would and wouldn’t do? And who bears the loss if it went further?
The law that answers them is spread across agency law, contract law, electronic transactions statutes, payments rules, computer-access law, AI regulation, and technical standards. Agent Authority puts it in one place, organized around those questions, with a short answer first and the sources behind it.
The site covers software acting for companies in their dealings with other businesses, under U.S. law. It doesn’t cover agents that deal with consumers, or consumer protection law.
Who’s behind it
Agent Authority was created and is maintained by Obehi Ota, a technology transactions attorney with almost two decades of legal experience. Her career spans corporate practice at Simpson Thacher, technology transactions roles at IBM and Wilson Sonsini, and independent practice. She earned her J.D. from the University of Pennsylvania Law School and holds degrees in Computer Information Systems and Information Science. She is licensed to practice law in New York and California. Obehi advises companies on agent authority through her law practice, Obehi Ota Legal.
How it’s researched
- Primary sources first. Statements rest on statutes, regulations, restatements and decisions, cited at the end of each page. Citations link to a free official or public copy where one exists. Restatements aren’t free to read online. Where the American Law Institute has a page for the volume, the link goes there. Commentary is used for context, not as authority.
- The law and the system together. Policies and contracts say what an agent is allowed to do. They can’t show what it can actually do. So the guides also ask which account the agent uses, what it can reach, where a person approves and what gets recorded.
- Settled law kept apart from open questions. Where the law is unsettled, pages say so and set out the competing views.
- Both views of AI agents. As of November 2026, no U.S. decision has been identified that resolves whether agency principles govern an AI agent’s contracts, or whether the agent is simply a tool, so the analysis considers both. See Tool or Agent?
- Both sides of the agent relationship. The guides set out the positions of the company using the agent and the company that provides it, and, where it matters, the seller, platform or customer the agent deals with.
- U.S. focus. Foreign court decisions appear only as analogues, and are labeled that way. Foreign regulation, such as the EU AI Act, is covered where it applies to companies that do business abroad.
Contact
For corrections, comments or anything else, please email [email protected]. Emailing doesn’t create an attorney-client relationship, so please don’t send confidential information.
This site provides general information about the law. It isn’t legal advice, and reading it doesn’t create an attorney-client relationship. How the law applies depends on specific facts and jurisdictions. Consult a lawyer about your situation.