United Airlines has decided that artificial intelligence has no business booking flights for its passengers and is shutting out third-party AI travel agents.
According to TechCrunch, the American airline announced that it would restrict automated assistants from accessing its booking systems, warning that allowing software bots to handle reservations could expose customers to security risks and potentially breach aviation regulations.
The decision is a kick in the teeth for technology companies promising that AI agents will soon organise our lives, make purchases and handle tedious administrative jobs while their owners concentrate on more important activities. United apparently thinks letting a chatbot loose with someone’s credit card and frequent-flyer miles is asking for trouble.
The airline says customers who share their login credentials with third-party AI assistants risk exposing sensitive information. An enthusiastic digital travel agent could also make an unwanted booking, change an existing reservation, or spend MileagePlus rewards without the owner’s permission.
United is particularly concerned about systems that operate autonomously after being given access to customer credentials. These tools can interact with websites and applications on users’ behalf, making decisions and completing transactions with limited human supervision.
This creates an awkward situation for companies promoting AI agents as replacements for ordinary website interactions. An assistant might find the cheapest flight, but giving it the authority to buy tickets raises questions about who is responsible when something goes wrong.
United has another objection involving aviation regulations. Under US Federal Aviation Administration and Department of Transportation requirements, passengers must personally acknowledge certain safety declarations during check-in, including confirming that their baggage does not contain prohibited substances.
An AI assistant completing the process automatically could bypass that requirement. The prospect of a chatbot cheerfully assuring an airline that its owner’s luggage contains no explosives is probably enough to give a compliance officer a nervous breakdown.
The restrictions will extend beyond United’s own website and mobile applications. The airline intends to modify its booking, ticketing and distribution policies to prevent unauthorised AI agents from gaining access through travel agencies and other intermediaries.
United said it would continue allowing customers to manage their journeys through its official application, which already supports booking flights, selecting seats and changing reservations. Presumably, the airline feels much more comfortable when the software making decisions belongs to United itself.
The company is joining a growing collection of businesses becoming suspicious of autonomous assistants. Banks have raised similar concerns about customers handing their credentials to third-party software, with Bank of America warning users against sharing passwords with AI agents.
The underlying security concern is straightforward. Once customers give an automated assistant access to an online service, they are trusting its developer to protect their credentials and ensure the software behaves appropriately. A compromised service or badly designed agent could potentially expose personal information or perform transactions the customer never intended.
Airlines have spent years developing systems that keep customers inside their own applications, where they can control the booking experience and sell additional services. Independent AI agents threaten to put another company between the airline and its passengers.
United has not suggested that every automated booking tool presents the same level of risk, and its announcement specifically targets unauthorised third-party personal assistants. Established travel distribution systems and approved commercial integrations are not necessarily affected in the same way.
The airline says it will update its Digital Trust Centre to explain the restrictions and guide customers whose automated assistants take unwanted actions. It plans to review its position as AI technology develops, although there is no indication of when it might relax the current restrictions.
For now, customers hoping to send a chatbot to book their next holiday will have to find another way to deal with United. The airline appears determined that when someone books a ticket, an identifiable human should be responsible for the transaction.
