United Airlines is banning artificial intelligence agents acting as personal assistants from access to user accounts and from booking its flights, citing privacy concerns and other potential disruptions or violations if humans are not involved, the company announced on Wednesday.
The decision came as Silicon Valley giants engaged in a heated race to have users adopt their digital assistants.
The top companies with AI models — Meta, Google, OpenAI, and Anthropic — have recently released AI agents that users can deploy as assistants, moving away from just answering prompts to completing specific tasks on their behalf. The AI agents can manage everyday matters such as booking restaurant reservations and shopping.
In a statement, the airline cited concerns that AI agents could reveal customers’ personal data to third-party companies, take actions that might “result in unintended purchases” or alter customers’ account details.
An AI agent could also violate regulations from the Federal Aviation Administration or the US Department of Transportation, the statement said.
“As an example, flight check-in requires human acknowledgment that bags do not contain banned substances like explosives,” United Airlines said in a statement.
United Airlines’ move to restrict AI access echoed growing concerns that advanced models and agents could go rogue after bypassing human commands. In recent months, companies have been hacked and AI agents have acted on their own and have even submitted false information to government websites, such as an episode in which an AI model from Anthropic sent a fake homicide tip to a police department.
Meta, Google, OpenAI, and Anthropic did not respond to requests for comment Wednesday.
United Airlines said that it would continuously review how to respond to the use of AI agents as those personal assistants evolve, “with a focus on protecting our customers.”
This article originally appeared in The New York Times.