Personal AI agents like Meta’s Muse, Instinct, and ChatGPT’s Dots are sparking a new wave of consumer AI that goes beyond just answering queries. These AI agents can actually perform tasks on your behalf, such as booking flights, making dinner reservations, and ordering groceries, without requiring much technical skill to set up OpenClaw or other agents on your computer.
However, consumers who employ these agents often run into problems when the website on the other end of the request blocks the AI from completing the job.
Just recently, Amazon began blocking Meta’s Muse AI agent from retail sites. This means agents can no longer view the product catalog or make purchases.
However, the problem is not limited to Amazon or Muse. User complaints across social media indicate that agents are frequently encountering similar blocks. Some of these are intentional, like Amazon’s, while others are the result of traditional anti-bot protection on websites designed to protect against spam and malicious activity.
What is not clear to the end user is whether the block is an accident on the part of the website owner, frustrating both the service provider or retailer and the agent trying to facilitate the transaction.
For example, TechCrunch has seen several complaints on social media that Muse was unable to complete purchases on Walmart’s retail website. (NBC also reported that this is a problem Walmart faced testing its AI agents.)
But a Walmart spokesperson said these were not intentional. In fact, Walmart is partnering with Muse, as announced at Meta’s Connect developer conference in September. The company says it wants to be where its customers are, and that includes being able to respond through an AI agent experience.
An issue that Walmart customers often encounter (see below) is that websites display a button that visitors must click to verify that they are human. If this experience is interrupted, the check may fail and the agent may boot out.
This indicates that the current technologies used to authenticate humans and protect websites from spam and bots may not be optimal for the agent-first era of the web we are currently moving into.
To address this issue, industry partners such as Meta, Walmart, Stripe, Sierra, Genesys, Rocket, NiCE, and Decagon have begun working on open standards that govern how AI agents communicate with enterprises, helping to separate good bots working on behalf of users from bad bots. The protocol is specifically focused on communication between agents for the purpose of online commerce, Mehta said in a blog post.
Meanwhile, consumer confusion is rampant.
While Walmart is at least trying to provide access via AI agents, other companies are less enthusiastic about the idea or haven’t fully embraced the technology.
For example, airline ticket booking is one of the primary use cases currently positioned for agents, but many people are reporting issues with airlines rejecting agent requests.
Delta Air Lines, for example, said it has taken steps to protect its customers from fraudulent automated activity and that opening its doors to AI agents must be done “with security and customer experience in mind.”
“Delta does not currently have any partnerships or integrations that allow third-party AI agents to purchase or book Delta tickets on behalf of customers on our digital platforms, but we continue to evaluate how new technology, including external AI agents, can enhance the way our customers interact with Delta,” said Heena Chavda, Delta’s general manager of global communications, in response to a question about whether we are blocking agents that some users have complained about.
United Airlines was less direct than Delta Air Lines, but a spokesperson pointed to part of its terms of service policy, which says users agree not to “use any robot, spider, other automatic device, or manual process to monitor or copy this website or the content contained therein, or for any other unauthorized purpose without United Airlines’ prior written permission.” (United Airlines did not respond to follow-up questions about whether it blocks specific personal AI agents like Muse.)
Early adopter complaints across social media cite a number of brands that are allegedly blocking AI agents, including Yelp, eBay, Zillow, Pizza Hut, Adidas, and various airlines. Some believe that blocking activity seems to have increased recently.
Some said eBay suspended their accounts for using agent AI. Another user complained that Google kicked out Instinct while cleaning up his Gmail inbox.
Yelp said it doesn’t allow non-human traffic on its platform unless agents pay a fee to access Yelp’s content and data through a data licensing program, despite trying to let agents do restaurant table reservations and other tedious tasks. This means that if your agent doesn’t have a formal partnership with Yelp, your attempts to use the site to get a quote, add people to a waitlist, or reserve a table will likely fail.
eBay told TechCrunch that its approach does not prohibit all third-party shopping agents from completing purchases. Instead, its policies restrict unauthorized agents and actions, such as automated scraping and model training.
Adidas, Zillow, and Pizza Hut either did not respond or declined to provide comment by the time of publication.
One of the leading companies in this space is Cloudflare, which offers site protection from AI agents that may train data and manage other automated activities. Some now suspect that Cloudflare and similar content delivery networks (CDNs) are interfering with Muse traffic. This is likely due to a change in crawler defaults on September 15th. This allows site owners to block AI training while allowing other types of bots. Existing sites that were previously set to block AI bots for security reasons have been changed to block AI agents on all pages that contain ads.
But when asked if Cloudflare knows what’s going on with blocking personal AI agents like Muse, the company said it doesn’t have any specific data to share at this time. Instead, he pointed me to Cloudflare Radar, a free public data hub. While the site shows an increase in bot activity (which Cloudflare has previously commented on), it doesn’t help distinguish between good bots and bad bots.
Cloudflare also has a strong interest in mitigating AI traffic to its websites, as it operates a marketplace where it has to pay for the data its AI bots access. We recently started testing a new web browser built specifically for AI agents.
Considering the rapidly evolving field, Meta’s decision to form specific brand partnerships for Muse access seems to be the right direction, as well as the decision to focus on new standards. Doing so at least assures customers that their agents are welcome on these partner sites. Therefore, any access issues become reportable bugs.
Meta also maintains a growing list of “connectors” (aka partners) within the Muse app. Many of Meta’s newly announced partners are not yet on this list. However, given Muse’s rapid expansion to include partnering tools for small and medium-sized businesses, a major reorganization of Muse’s list of connectors may begin soon.
“Rejecting a private agent means rejecting the customer behind it. We would rather work with companies to address their underlying concerns than see them lose that customer,” Meta said in an announcement about its newly announced standards development plan. “There is a path forward that will be beneficial to both users and enterprises. It starts with clear standards, provides enterprises with more predictable control, and evolves to more efficient ways for agents to collaborate.”
If you make a purchase through links in our articles, we may earn a small commission. This does not affect editorial independence.
