AI agents face website blocks as businesses weigh how to respond

Personal AI agents such as Meta's Muse and Instinct or ChatGPT's Dots can book flights, make restaurant reservations and order groceries for users. But people using them increasingly find that a website blocks the agent, leaving the task unfinished.

Deliberate and accidental blocks

Amazon recently began blocking Meta's Muse from its retail site, so the agent can no longer browse the catalog or make purchases. The issue goes beyond Amazon, though. Some blocks are intentional, while others stem from traditional anti-bot tools meant to guard against spam and malicious activity.

Walmart said Muse's problems on its site are not intentional, as the company is a Meta partner. The difficulty is a button that must be clicked to prove the visitor is human: if the process is interrupted, the check fails and the agent is kicked out.

How other companies respond

Delta said it has no partnership allowing third-party AI agents to book its flights and is protecting customers from unauthorized automated activity. United pointed to its terms of use, which bar robots and other automatic devices without written permission. Yelp allows non-human traffic only when an agent pays for access through its data licensing program. eBay said it does not prohibit all third-party shopping agents but restricts unauthorized ones and actions such as scraping and model training.

Cloudflare changed its crawler defaults on September 15, and sites previously set to block AI bots began blocking AI agents on pages with ads. The company said it had no specific data on the blocking of Muse.

A shared standard

Meta, Walmart, Stripe, Sierra, Genesys, Rocket, NiCE and Decagon have begun work on an open standard governing how AI agents communicate with businesses, helping to tell good bots from bad ones. The protocol will focus on agent-to-agent communication for online commerce. Meta argues that turning away a personal agent means turning away the customer behind it.