United Airlines blocks AI agents from booking flights

United Airlines has decided not to let customers use AI agents to book flights or manage reservations, citing cybersecurity risks. The policy is a notable setback for the vision of autonomous AI assistants that can act on behalf of users across the web. In that vision, an agent might compare fares, complete a booking, change a seat, or handle a cancellation without the customer navigating every page. United's stance shows that companies are not ready to grant that kind of access when sensitive accounts and payments are involved. Airlines handle payment data, loyalty accounts, identity information, and travel records. They worry that third-party agents could be manipulated, poorly secured, or used to make unauthorized changes. A booking agent that acts for a customer also needs a reliable way to prove authorization and limit what it can do. Without those safeguards, the airline may see more fraud, account takeover, or customer disputes. United's decision may encourage other companies to set their own rules for agent access rather than waiting for a common standard. It also points to a larger challenge for the AI assistant industry: autonomy is easy to demonstrate but hard to secure. For AI agents to move from demos to everyday transactions, they will need clear identity, permission, and audit systems. Businesses will want to know who is acting, what they are allowed to do, and how to reverse mistakes. United's policy is a warning that the agent economy will be built on trust and access control, not just better models.