United Airlines will block AI agents from booking flights through its website, third-party travel agencies, and flight aggregators, the carrier announced on November 11. United cites security, privacy, and federal safety rules as its reasons.
What United says the risk is
United’s announcement says handing payment details or account logins to an AI agent could lead to unintended purchases, account changes, or misuse of MileagePlus miles. The airline also points to a regulatory concern: letting a software agent complete a booking could conflict with Department of Transportation and Federal Aviation Administration rules that require a human to confirm certain safety details at check-in, such as confirming a checked bag contains no explosives.
The policy applies across every channel United does not fully control itself, meaning it covers bookings attempted through its own site by an automated agent, through online travel agencies, and through flight-search aggregators that compare fares across carriers. United has not said how it will technically detect and block agent traffic, only that the restriction now applies company-wide.
United’s CEO has been publicly wary of AI booking agents
United chief executive Scott Kirby has previously said he does not want to hand control of bookings to a third-party AI layer. Speaking at an industry event on November 10, Kirby said: “To be honest, I don’t want to create another middleman.” The Wall Street Journal has reported that Kirby has been reluctant to embrace agentic commerce more broadly, a stance that this new policy makes explicit.
Airlines and major platforms are splitting on AI agent access
United is not alone in restricting AI agents from transacting on its behalf. JPMorgan Chase and Amazon have both taken similar positions, blocking automated agents from completing purchases or financial transactions on their platforms. Southwest Airlines has gone the other way, launching a plugin that lets customers book flights directly through ChatGPT.
| Company | Position on AI agent bookings/transactions |
|---|---|
| United Airlines | Blocks AI agents from booking, citing security and FAA/DOT rules |
| Southwest Airlines | Launched a ChatGPT booking plugin |
| JPMorgan Chase | Restricts AI agents from completing transactions |
| Amazon | Restricts AI agents from completing purchases |
How much are AI agents actually booking travel right now?
Uber’s own Q3 earnings call offers a rare real-world data point on how much agentic commerce is actually happening today. The company reported 320 million trips and deliveries in the week around Halloween, of which AI agents accounted for about 500 Uber Eats orders and four rides. That is a vanishingly small share of total volume, which suggests United’s new rule is aimed more at closing a door before it opens wider than at a problem already straining its systems.
Why this matters beyond one airline’s policy
Airlines sell a product where a single mistaken booking, a wrong date, a missed connection, a mishandled special-assistance request, carries a higher cost to fix than a mis-sent parcel. That is part of why United is drawing its line specifically at completed transactions rather than at AI tools in general. The airline already uses AI elsewhere in its operations, including automated rebooking and weather-delay alerts, but has kept a human or a direct, owned channel in control of the final purchase step.
The policy also puts pressure on the fast-growing field of AI shopping agents built by OpenAI, Google, and smaller travel-tech startups, which depend on merchants being willing to let them complete purchases on a user’s behalf. If more airlines follow United rather than Southwest, agent developers may find travel booking becomes one of the harder categories to automate end to end, even as it becomes easier to automate research and comparison.
Frequently asked questions
Can I still use a travel app’s AI assistant to search for United flights?
United’s policy targets agents that complete a purchase on a traveler’s behalf, not tools that merely search or compare fares. The restriction is on booking, not browsing.
Does this affect United’s own AI features?
No. The restriction is aimed at third-party AI agents acting through United’s website, travel agencies, or aggregators, not at tools United itself builds and controls directly.
Why do safety checks require a human specifically?
United points to Department of Transportation and Federal Aviation Administration requirements built around a human traveler confirming specific safety declarations, such as verifying that checked baggage contains no hazardous or explosive material, at the point of check-in. Those rules predate agentic AI and were not written with a software intermediary in mind, which United says creates compliance risk if an agent books on a traveler’s behalf without that direct confirmation ever happening.
Leave a Reply