The Pulse
Websites Block AI Agents From Shopping, Booking and Banking
Users report that AI agents are being blocked or tripped up by security systems on popular websites, while some companies restrict or have not enabled agent access. Cloudflare’s new controls let site owners set separate permissions for sear
AI.info Team · Written with AI ·
Four agent services, and familiar sites that say no
Users of at least four AI-agent products—Meta’s Muse, xAI’s Grok, OpenAI’s Dots and Instinct—have reported that popular websites blocked them from completing tasks, according to The Verge’s report published October 17, 2026. The reported obstacles range from restrictions aimed specifically at AI to security systems that may mistake automated activity for abuse.
Some companies have confirmed deliberate limits or said they have not enabled agent access. Yelp told TechCrunch that it allows agents only through its data-licensing program. Delta Air Lines said it has no partnership or integration that enables third-party agents to book flights on its digital platforms, while saying it is evaluating the technology; that statement did not declare a blanket ban. JPMorgan Chase and Charles Schwab have confirmed blocking agents from account access because of security concerns, The Verge reported. Users have also said they can no longer reach Cars.com and Amtrak through agents, though those reports do not establish whether the companies imposed a formal ban.
The distinction matters: a failed task does not necessarily mean a site has adopted a policy against AI. Automated services may trip defenses designed to stop scams, hacking or other suspicious activity. Chris Canal, CEO of the AI safety evaluator EquiStamp, told The Verge:
“One of the claims is that they’re going to permeate every aspect of our lives and be able to do anything that we could do online.”
Chris Canal, CEO of EquiStamp, told The Verge.
Amazon’s fight reaches beyond Perplexity
Amazon has become the clearest flashpoint. After banning Perplexity’s agentic shopping feature and suing the startup last year, the retailer also challenged Meta’s Muse, saying its “continued access” violates Amazon’s site policies. Users have also reported trouble using Grok, Instinct and OpenAI’s Dots on Amazon.
The dispute is not only about whether an agent can reach a page. It concerns who controls automated purchases and what a retailer loses when software, rather than a person, browses its store. Perplexity has argued that its agent acts as an extension of the user; Amazon has pressed its objections through litigation. The conflict raises questions about the access agents from other companies may have to websites.
Meta has taken a different route, partnering with retailers including Walmart, Gap, Sephora and Best Buy so Muse can shop on their sites. Shopify has also opened its network of storefronts to agents. The Verge says it sought comment from OpenAI and Google; Instinct declined to comment.
Cloudflare separates search from agent access
Infrastructure providers are giving site owners more ways to set those boundaries. Cloudflare announced on September 15, 2026, that it was replacing its single “Block AI Bots” switch with separate controls for search, AI training and AI agents. For new websites on Cloudflare, its proposed recommended settings leave search crawling on, disallow AI training and block agents on pages carrying ads. For other new sites, the recommendations allow all three. Customers can change those settings.
The split recognizes that website owners may want their pages to appear in search results while refusing training or automated transactions. It also shows why “AI access” is not one decision: a search crawler, a system collecting material for model training and an agent acting on a customer’s behalf can have different permissions. The reported blocks and Cloudflare’s new settings reflect a more finely divided set of choices, not a single, measured count of websites turning agents away.
Human users may face the extra steps
Broad anti-bot defenses can catch people along with automated software. The Verge reported complaints that some Google accounts used with Muse were blocked for “suspicious activity,” and that some users had trouble checking out on Walmart despite the retailer’s partnership with Meta. Those reports illustrate how a site’s defenses can create friction even when it has agreed to work with an agent.
Canal warned that more sites may ask visitors to prove they are human, including during ordinary online shopping:
“We could see them on a simple grocery shopping website soon, where you’re having to jump through a bunch of hoops to prove that you’re a human using the computer or your phone.”
Chris Canal, CEO of EquiStamp, told The Verge.
For now, the reporting documents specific blocks and user complaints, but offers no estimate of how many websites prevent agents from operating or how often defenses mistakenly stop human customers.