Amazon, United, Target: The Companies Now Banning AI Agents

United Airlines announced on Wednesday that unauthorized AI agents are barred from searching, booking, buying, or managing trips on its platforms, making it the latest major company to write bots out of its terms of service. The reason is blunt: a bot checking in on a flyer's behalf could violate federal regulations that require a human to acknowledge their bags contain no banned items, such as explosives.

United joins Amazon, Booking.com, eBay, Resy, Target, and Yelp in restricting automated assistants. The wave began after Meta released its Muse assistant in September, with similar tools like Instinct following, and mainstream chatbots like ChatGPT and Claude adding agentic features. The bans cover two behaviors: agents that buy directly on a user's behalf, and in some cases agents that browse a site at all.

The companies cite privacy and security concerns first. But there's a second motive in play. If humans stop visiting sites directly to browse, buy, and book, those sites lose the traffic that funds them.

What does Amazon's ban actually say?

GeekWire first reported in September that shoppers trying to use Muse on Amazon saw a popup: "continued access by an unauthorized AI agent violates Amazon's Conditions of Use, to which our customers have agreed."

The conditions don't name every AI agent explicitly. They give Amazon the right to decide case by case, stating that "at our sole discretion, we may limit, including by technical measures, whether and how any Agent accesses, uses, and interacts with Amazon Services."

Who blocks first and asks questions later?

Booking.com takes the hardest line of the group. Its terms prohibit AI-powered assistants from making bookings or reservations without express permission, and from accessing the platform for any reason without prior authorization. Enforcement is direct: the company says it will block anyone it suspects of using an AI assistant.

eBay updated its user agreement to prohibit automated access without permission, and it names the specific offenders: "buy-for-me agents," "LLM-driven bots," and any "end-to-end flow that attempts to place orders without human review."

Yelp bans non-human traffic without express permission too, but it left a paid door open. Its data-licensing program allows approved integrations to access data and take certain actions, including one that lets ChatGPT users make restaurant reservations.

Which agents made the approved list?

Two companies chose whitelisting over blanket bans.

Target's terms prohibit AI shopping agents from accessing or gathering data from its site unless explicitly authorized. The approved list includes "Target's AI Shopping Assistant, OAI, Google and Microsoft." Approved agents can sign in to accounts, build carts, place orders, and initiate returns with customer permission. Target reserves the right to change the list or restrict agents' access at any time.

Resy, the restaurant booking platform, supports discovery and booking through ChatGPT and Claude, but can block AI activity outside those approved services. The company cited platform security and excessive demand on its system, and said users who violate the rules could have their accounts terminated.

Why does the split matter?

The seven policies reveal two distinct strategies. Amazon, Booking.com, eBay, and Yelp treat agents as unauthorized by default and gate access behind permission, contracts, or payment. Target and Resy treat agents as a controlled channel, letting a handful of vetted bots transact while blocking the rest.

The United case shows a third pressure: regulation. When a federal rule requires a human acknowledgment, an autonomous agent can't lawfully complete the step, no matter how good the terms of service are.

Meanwhile, the divide isn't universal. Walmart, by contrast, has been adding AI-agent shopping features even as Amazon blocks bots, a split that suggests the industry's answer to autonomous shoppers is still being negotiated platform by platform.

With Muse, Instinct, ChatGPT, and Claude all pushing agentic capabilities into everyday tasks, expect the whitelist-versus-ban divide to define which assistants survive on major commerce and travel sites, and which hit a popup instead.