United Airlines has updated its terms to restrict unauthorised third-party AI agents on its website and app, while explicitly exempting search engines that index its public pages. The restriction lives in the terms, not in robots.txt.
United Airlines updated its website terms and contract of carriage on November 11, 2026, restricting unauthorised third-party AI agents on its site, app, and through travel agencies and booking sites. Any AI agent that searches, books, buys, manages, or collects information on united.com or the United app needs United's prior written permission. The restriction covers the full range of what an agent might do on behalf of a traveller, from looking up a flight to completing a purchase.
United reserves the right to block such agents at any time, even when the traveller authorised them. That last point matters: the traveller's permission is not enough, and United can cut an agent off regardless of what the customer asked it to do. United says it does not yet work with any third-party AI agents, so the new terms are not ending an existing partnership; they are drawing a line before the partnerships begin, and making clear that any agent access will happen on United's terms or not at all.
There is one reassurance for travellers: bookings agents made before November 11 will not be cancelled. The change is not retroactive for people who already used an agent to book, which avoids punishing customers for something that was allowed when they did it. But from November 11 onward, an unauthorised agent booking is a breach of United's terms.
Search engines indexing United's public pages are explicitly exempt from the restriction. United is not blocking Google or other search engines from crawling its site, and the terms draw a clear line between a search engine showing United's pages in results and an AI agent acting on a traveller's behalf. That distinction is important for SEOs: United's pages can still be indexed, ranked, and found through search, even as agent access is shut down.
The terms also bar using United's fares, schedules, availability, or pricing data to "train, fine-tune, ground, evaluate, or otherwise develop" an AI model. That is a broad ban, covering not just using the data to book a flight but using it to build or improve an AI system at all. United is protecting its data from becoming training material for models that might later compete with it or reproduce its pricing.
As of November 14, United's robots.txt named no AI crawlers or agents. The restriction lives in the terms, not robots.txt, which means it is a legal restriction rather than a technical one. An AI crawler that respects robots.txt would not be stopped by United's robots file, because there is nothing in it to stop it. The enforcement, if it comes, will be through the terms and through blocking, not through the crawling protocol.
United cited privacy and account-security risks, unintended transactions, and FAA and Department of Transportation compliance concerns. Those are real risks for an airline: an agent that books the wrong flight, accesses an account without proper authorisation, or triggers a transaction the traveller did not intend creates problems that are hard to undo. But the compliance angle is the one to watch, because regulated industries have reasons to restrict agents that go beyond protecting their own data.
For SEOs, United's move is a useful case study in how companies are drawing lines around AI access. The pattern emerging across industries is a split: search indexing stays open, because companies want to be found, while agent access and model training are restricted, because companies do not want their data used to act or to train without permission. That split is likely to become the standard position, and SEOs should expect to see it in more terms of service.
For AI builders, the lesson is to check the terms, not just robots.txt. A site with an open robots file may still forbid agent access in its terms, and a company that welcomes search crawlers may still block an agent that books on a user's behalf. United's terms are now the document that matters, and anyone building travel agents, booking tools, or shopping assistants needs to treat terms of service as a data source in their own right, alongside the crawling signals they already check.
Sources: Search Engine Journal, November 14, 2026.
Related reading on this site: How Often Sites Block AI Crawlers: 14,318 Sites Measured, Fake Googlebots Are Passing Reverse DNS Checks, Log Study Finds, Google: Serve a 503 If You Go Offline to Patch PHP, OpenAI's Aura Safety Check Can Block Your Site From ChatGPT Browsing, Cloudflare Signals Helped OpenAI Find Hundreds of Millions of New Pages, and Bing Adds a -ai Operator and Browser Extension to Turn AI Answers Off.
Linkbaits helps you create content that earns citations and backlinks. Start with the toolkit and build your first linkable asset this week.
Start free trial