The gate, and the assistant in your hand.
By Ellis Ward | November 14, 2026
United Stands Alone on AI Agent Bookings While Rivals Shift Risk to Travelers
Every major U.S. airline had to decide what to do when AI agents started booking flights. Only one said no.
That carrier is United Airlines. Forbes reported on November 12, 2026, that United is the only major U.S. carrier actively blocking third-party AI agents from its booking systems. Its rivals have taken the opposite position: they accept agent-initiated bookings and tell customers that if something goes wrong, the customer bears the cost.
SC Media confirmed the restriction on November 13, reporting that United Airlines has restricted third-party AI agents from its booking systems. NJ.com covered the story the same day, its headline asking and promising to answer why United made the call. Neither outlet identified another major carrier that had followed United's lead.
The Forbes headline put the split plainly: "United Airlines Alone In Blocking AI Agents From Booking Flights - Rivals Say Customers Assume Risk." That framing captures two distinct positions. United drew a line at the booking interface. Its competitors drew theirs further down, placing the exposure on the traveler who used the agent.
Rivals' "Customers Assume Risk" Position Has Real Consequences
When an airline tells travelers that AI-agent bookings happen at their own risk, the airline is not offering helpful information. It is repositioning legal exposure.
RFTUV's November 12 piece on United's ban raised the question that has not gone away: when an AI agent makes a booking error, who pays? Three parties sit in the frame. The airline took the booking and issued the ticket. The AI agent maker built and deployed the product that executed the transaction. The traveler authorized the agent to act and may have done nothing further to verify the result before the ticket went through.
Competitors that accept AI agent bookings appear to have resolved this question informally, in their own favor. They let the agent book, and, as Forbes framed it, their position is that the customer who chose to delegate the task assumes the risk.
Whether that position holds in an actual dispute is not settled. Courts have not ruled on AI agent liability in consumer travel transactions. The existing frameworks for consumer protection assume a human made a purchasing decision, confirmed the purchase, and stands ready to dispute any error. An AI agent that completes all three of those steps without a human review step does not fit cleanly into any of those assumptions.
A Disclosure Problem Travelers May Not See Coming
There is a practical gap in how this plays out for travelers. When a traveler's AI agent books a flight on a carrier that accepts agent-initiated bookings, the carrier's risk disclosure does not appear in the agent's interface. It lives in the carrier's terms of service. The agent books, the ticket is issued, and the traveler may not know the carrier has defined this transaction as the traveler's own problem until there is a problem to solve.
If the agent books the wrong date, selects the wrong cabin class, or fails to apply a loyalty account, the traveler holds the ticket and the error. The carrier holds the revenue and an argument.
That argument may be a strong one. It may not be.
United's Reasoning Stays Partly Closed
The NJ.com report from November 13 addressed the question of why United made this call. United has not issued a detailed public statement laying out the technical or legal reasoning in full. The reporting confirms United has restricted third-party AI agents from its booking systems specifically, the restriction targets the booking interface, and no other identified major carrier has taken the same position.
The restriction applies at the transaction layer of travel. AI agents booking flights are not reading schedules or checking fares. They are creating records: flight purchases, seat selections, itinerary changes. Those records carry the same weight as ones a human traveler generated directly, and they carry that weight whether or not a human reviewed the agent's work before it went through.
United's restriction reads as a refusal to be on the receiving end of a transaction it cannot confirm was human-initiated. Whether that is a legal posture, a technical constraint, or both is not resolved by the reporting available.
The Risk Has Not Been Neutralized, Only Moved
United's ban did not remove the AI agent liability question from travel. It moved the question to competing carriers.
Travelers whose AI tools simply reroute around United to a carrier that accepts agent bookings have not avoided the problem. They may have stepped further into it. The carriers accepting those bookings are, on the Forbes account, treating those bookings as the traveler's own risk. The traveler, in many cases, has not read those terms before the agent acted on their behalf.
The AI agent makers have not publicly addressed what happens when their products produce a booking error on a carrier that has disclaimed responsibility. No regulator has entered that gap. The carriers have stated their position. The agent makers have not. The travelers in the middle have no settled recourse to point to.
The industry's informal test of AI agent bookings is running. United chose not to participate. The question of what happens when that test produces a costly failure for a real traveler has no answer yet.
RFTUV first reported on United's restriction on November 12. This update reflects trade coverage published November 12 and 13, 2026, by Forbes, SC Media, and NJ.com.
