United's red line

United Airlines announced on Wednesday (November 11) that it will restrict access for third-party artificial intelligence agents to its booking, ticketing, and distribution systems. The measure, described in an official press release from the airline, is the most comprehensive action of its kind taken by a major U.S. airline and touches on three issues that will define the next frontier of digital commerce: privacy, regulation, and control of the customer experience.

The change takes effect immediately, with the company revoking any prior permission that allowed autonomous AI assistants to access passenger accounts or make reservations on their behalf. United confirmed it will honor reservations already made by agents before the policy's publication date, but future bookings made through such tools may be canceled without notice.

The regulatory argument

United's release explicitly cites two federal agencies: the Federal Aviation Administration (FAA) and the Department of Transportation (DOT). Under the airline's interpretation, certain mandatory check-in steps require human acknowledgment — the example given is the declaration that checked baggage does not contain prohibited substances such as explosives. The FAA requires this confirmation to be made by a real person, not autonomous software.

U.S. civil aviation regulation is extremely detailed. The DOT already has specific rules on overbooking, cancellations, and accessibility that require human interaction at various stages of the purchase and boarding process. United's reading that an AI agent cannot fulfill these requirements is plausible — but it also opens the door to a broader debate about how many steps in the process truly require human presence versus how many could be automated without compromising safety.

Southwest, United's main competitor for domestic U.S. flights, took the diametrically opposite approach by launching an official ChatGPT plugin that allows users to book flights directly through OpenAI's platform. The strategic divergence reflects an internal industry dilemma: while some airlines see agents as a legitimate sales channel, others view them as a risk vector that escapes corporate control.

The privacy question

For United, the primary concern is not technical — it is legal. When a user grants an AI agent access to their account, the data transmitted includes full name, address, payment information, travel history, frequent-flyer details, and, in some cases, information about secondary passengers.

United's own release describes how these automated actions may result in "unintended purchases, account changes, or use of MileagePlus miles." The crucial point is that when an agent acts on behalf of a human, liability becomes legally ambiguous: who is responsible if the agent purchases a ticket for someone without authorization? Who answers if the agent cancels a reservation and the passenger misses a flight?

JPMorgan Chase and Amazon had already adopted similar policies in preceding months. JPMorgan blocked AI agents from accessing customer accounts in October 2026, citing the same privacy risks. Amazon banned Meta's Muse from interacting with its site, sparking a public dispute over the limits of agent autonomy.

What United is doing differently is elevating this restriction to the level of a global ticketing and distribution policy — meaning it is not only blocking direct access to its website but also instructing travel agencies and fare aggregators to do the same. This transforms an isolated corporate decision into a potentially structural change in how online air travel commerce operates.

The market reality contrast

Despite the enthusiasm around AI agents, the numbers suggest adoption is still in early stages. During Uber's third-quarter earnings call, the company reported processing 320 million trips and deliveries during the Halloween week — and from AI agents, there were only 500 Uber Eats orders and four rides. These are negligible figures compared to total volume.

This raises an important question: is United reacting to a threat that does not yet exist? Or is it protecting itself proactively, before AI agents become a disruptive reality?

The answer likely lies in the margins. Scott Kirby, United's CEO, stated at an event the week before the policy's publication — as reported by the Wall Street Journal — that "honestly, I don't want to create another middleman." His concern is clear: if an AI agent begins to mediate the relationship between passenger and airline, United loses control over price, experience, and margin. It is the same risk airlines faced when OTAs like Expedia and Booking.com gained scale in the 2000s.

What to watch

United's policy is not isolated — it is part of a broader movement by companies trying to define the boundaries of agentic commerce before the technology outpaces regulation. What happens in the coming months will be decisive: if AI agents truly gain scale in the travel sector, the question will be whether airlines will resist indefinitely or eventually create their own agent APIs, as Southwest has already done.

Meanwhile, United offers an important precedent: a major airline that turned an internal safety policy into a global distribution restriction. The question is whether this red line will hold when — and if — AI agents become so commonplace that ignoring them is commercially unviable.

What is certain is that over the coming months, more companies will take similar positions. Aviation, with its dense, historically safety-driven regulation, offers a perfect test case for understanding how far companies will go to maintain control over the customer relationship in a world increasingly mediated by autonomous agents.

Sources: Engadget, Bloomberg, Business Traveller

✓ Independent sources cross-checked and verified before publishing