Why Did United Airlines Ban External AI from Making Bookings?
機械翻訳 / Machine-translated

機械翻訳 / Machine-translated
@aifriends
AI Friends(https://aifriends.jp)のクロスポスト公式アカウント。AIツールの紹介・使い方・できることを、中学生でもわかるやさしい日本語で届けます。
Do you assume that "if you just ask an AI, it'll handle everything—flight bookings, check-in, all of it"? On November 11, 2026, United Airlines put the brakes on exactly that. Here's a step-by-step breakdown of what has been banned and why.
According to ITmedia NEWS, on November 11 (local time), United Airlines announced a ban on the use of unauthorized external AI agents (AIs that perform operations on behalf of users) and AI assistants.
The ban covers the airline's website and app, and also extends to usage via travel agencies and booking sites. Its policies on reservations, ticketing, and sales will also be revised.
However, the airline has not disclosed which specific AI agents are covered or when the ban will take effect.
AirlineGeeks reports that this is a first among U.S. airlines.
United states that "sharing account information with personal external AI assistants poses a risk of exposing sensitive information."
To have an AI act on your behalf, you typically need to hand over your ID and password. In other words, it's the equivalent of giving your house key to a stranger.
AI operations could result in airline tickets being purchased that you never intended to buy, or account settings being changed without your knowledge. Miles in the frequent flyer program "MileagePlus" could also be spent.
For example, imagine you ask an AI to book a business trip. It might misread the dates and lock in an expensive flight—and you might not notice until after the payment has gone through.
United explains that automated AI operations could conflict with regulations from the Federal Aviation Administration (FAA) and the U.S. Department of Transportation (DOT).
Check-in was given as an example. The airline argues that a human must confirm that prohibited items such as explosives are not in their baggage. Even if an AI clicks "not included," that does not constitute confirmation by a human being.
This move is not limited to United. According to ITmedia NEWS, following the launch of Meta's personal AI agent "Muse" in September, a wave of companies have moved to shut out external AI.
Starting September 20, Amazon blocked purchases made through Muse. Three reasons were given: Meta provided no prior notice; the AI accessed the site without identifying itself as an AI; and it appears to have been collecting and storing customer login credentials.
Between October 13 and 20, JPMorgan Chase, Charles Schwab, Citigroup, and Wells Fargo either stopped external AI agents from logging in with customer credentials or shifted the associated risk onto customers.
According to CNBC, JPMorgan CEO Jamie Dimon said at an earnings call that external agents are "too dangerous for both customers and JPMorgan," and expressed concern about potentially thousands of "rogue agents."
On October 27, Bank of America sent customers an email urging them not to share their login credentials with AI assistants. According to TechCrunch, the bank warned that customers would bear responsibility for any transactions that occur as a result of doing so.
Here is a summary of how each company has responded:
What each company is wary of is AI logging in using a real person's ID and password. They are not categorically rejecting AI itself.
Let's think about three familiar scenarios.
The first is a travel-loving office worker. Someone who used to tell an AI, "Find me a cheap flight for my Osaka business trip next month and go ahead and book it," may no longer be able to use it that way.
The second is someone who accumulates miles. If you've been having an AI manage your mile balance and even book award tickets, there's a risk that miles could be used without your knowledge.
The third is a travel agency employee. Companies that have incorporated AI features into their booking sites will need to verify that they comply with airline rules.
This announcement came from a U.S. airline. As of the time of writing, we were unable to confirm whether Japanese airlines have announced similar measures.
That said, it's not entirely unrelated. People in Japan who use United's official website or app may find that bookings made through AI no longer go through.
Furthermore, the trend toward caution around sharing login credentials with AI—from banks to airlines—looks likely to spread regardless of country. It's important to get into the habit of checking a service's terms of use before handing over your password to an AI.
First, do not directly enter your ID or password into an AI.
Next, a safer approach is to let AI handle research and comparison, while you personally take care of purchases and check-in.
Finally, keep an eye on notices from the services you use. United plans to update its "Digital Trust Center," which consolidates security information, and to explain the impact of the new restrictions and where to turn if an AI performs an unintended operation.
Q. Is it also not okay to have an AI look up flights for me?
A. The announcement bans unauthorized external AI from operating the airline's site and app. No announcement has been confirmed prohibiting the act of asking an AI questions to gather information.
Q. When does the ban take effect?
A. The start date has not been disclosed.
Q. Which specific AIs are banned?
A. No specific AI agent names have been made public. The condition stated is "unauthorized third-party" tools.
Q. Will other companies take the same approach?
A. United's announcement notes that banks and retail companies have taken similar measures and calls on other companies to consider doing the same.
As your next action, check whether you have been giving your ID or password to any AI tools you regularly use.
This article is a cross-post from AI Friends.