Skip to content
Bamboo Flags
Custom automation & AI for growing businessesDiscover · Design · Deliver · Drive

Websites are blocking AI agents: what this means for your automations and your own site

Amazon, Delta, Yelp and two US banks block AI agents that act for users. This article lists who blocks agents and why, what is not known, and the checks to do on your automations and on your own website.

Bamboo Flags3 min read

In September and October 2026, several large websites started to block AI agents that browse and buy for their users. TechCrunch reported on 6 October that users of AI agents such as Meta's Muse, ChatGPT's Dots and Instinct often cannot finish a task on these sites. The Verge added banks to the list on 17 October.

This affects a small business in two ways. An automation that uses an agent on another website can stop without warning. And your own website can block the agents of your customers.

Who blocks agents, and what they say

  • Amazon blocks Muse. According to The Verge, Amazon said that Muse's "continued access" breaks its site policies. TechCrunch says that Bloomberg first reported the block on 21 September. Users also report that agents from OpenAI and xAI cannot use Amazon.
  • Delta has no integration for third-party agents. A Delta spokesperson told TechCrunch that the airline protects customers from "unauthorized automation".
  • Yelp allows non-human traffic only if the agent's operator pays for its data licensing program.
  • United points to its Terms of Use. They forbid robots and other automated devices without written permission.
  • JPMorgan Chase and Charles Schwab confirmed to The Verge that they block agents from customer accounts. They give security as the reason.

Reports that the companies did not confirm

Users report blocks on eBay, Zillow, Pizza Hut, Adidas, Cars.com and Amtrak. These reports come from social media. Zillow told TechCrunch that it allows agents that act for consumers, but its anti-scraping tools can block some of them. Walmart, a Muse partner, said that failed Muse purchases on its site were not intentional.

Why blocks happen when nobody chose them

TechCrunch names one likely cause: the "verify you're human" check. When the agent cannot pass the check, the site stops the session.

Another possible cause is Cloudflare. On 15 September Cloudflare changed its default settings for crawlers. Site owners can now block AI training and still allow other bots. According to TechCrunch, some sites that already blocked AI bots for security now also block AI agents on pages with ads. Cloudflare had no data on personal agents specifically. If your website runs behind Cloudflare, these defaults can apply to it.

What the industry proposes

Meta works with retailers one by one. Its partners include Walmart, Best Buy, Gap and Sephora. Meta's position, as quoted by TechCrunch, is that "turning away a personal agent means turning away the customer behind it." According to The Verge, Shopify opened its network of storefronts to AI agents.

Meta, Walmart, Stripe and five other companies are also working on an open standard for how agents talk to businesses. No draft is public yet.

What is not known

  • No company has published how many agent sessions it blocks.
  • There is no standard way for a website to tell a good agent from a scraper. Cloudflare Radar shows bot traffic but does not separate the two.
  • The unconfirmed blocks can be deliberate or a side effect of bot protection. The reports do not show which.

Checks for your automations

  1. List every automation that uses an agent or a browser bot on a website you do not control.
  2. For each website, read the Terms of Use for rules on automated access.
  3. Prefer the official API or integration when one exists. An API does not depend on bot detection.
  4. Add an alert for when a run fails. Do not let a blocked step fail without a message.
  5. Do not give an agent access to a bank or broker account. JPMorgan Chase and Schwab already block this.

Checks for your own website

  1. If you use Cloudflare or another CDN, open the bot settings and read the current defaults for AI crawlers and AI agents.
  2. Decide which traffic you want: AI training crawlers, search crawlers, and agents that act for a customer are three different cases.
  3. Give an AI agent a real task on your website, for example a booking or a purchase. Make a note of where it stops.
  4. If the agent stops at a "verify you're human" check, decide if this is what you want. An agent that stops there cannot finish the order for its user.

Found something worth fixing?

Bring it to a free 45-minute call.

Book a free 45-minute call

More from the blog