Why Some Websites Are Shutting the Door on AI Shopping Helpers

Why Some Websites Are Shutting the Door on AI Shopping Helpers
Meta's Muse AI assistant ran into verification checks at Walmart, one of its retail partners. Amazon, by contrast, deliberately blocked Muse from shopping on its site. The episodes, reported by Engadget on 2 November 2026, sit at the center of a growing argument about who may let AI agents act for people online. That argument is now spreading from shops into banking.
An AI agent is a program that carries out tasks for you, such as finding a product, filling a basket or booking a flight.
Walmart said its verification checks were not meant to stop the Muse assistant. The checks still presented Muse with a barrier on a site where Meta has a partner relationship.
Amazon took the opposite approach.
Amazon's position goes beyond Muse. The company says Perplexity's Comet AI browser breaks its access rules and puts customer data at risk. Perplexity says users have authorized its Comet agent to act on their behalf.
The dispute has gone to court. In August 2026, a federal appeals court vacated a preliminary injunction against Perplexity. The underlying disagreement between Amazon and Perplexity is still unresolved.
Amazon also runs its own Buy for Me feature, which can buy products from third-party brands' websites. So the company restricts outside agents while operating one of its own.
Other merchants are writing limits into their terms. eBay's user agreement bans unauthorized buy-for-me agents unless eBay grants permission. Delta said it has no integration that lets third-party agents book flights through its website. It did not confirm whether it deliberately blocks specific AI agents.
Banks have been more direct about the technical risk. On JPMorgan Chase's October earnings call, CFO Jeremy Barnum said some AI agents use shared login credentials and screen scraping to move money. He called these methods "not secure." JPMorgan Chase blocks them where possible.
Barnum also warned that customers who hand over their credentials could be responsible if an AI agent takes an action they did not want.
According to Reuters, published on 22 October 2026, major U.S. banks have been blocking third-party AI agents from operating inside customer accounts and making transactions. Earlier, on 22 September 2026, Reuters reported that NatWest warned that using AI agents for online shopping could increase the risk of scams, fraud and data-privacy breaches.
Why this is hard to fix
The core problem is the same in every case. An agent that logs in with your password, or reads a web page and clicks through it the way a browser does, looks exactly like you to the website. It also gets whatever permissions your login carries.
Think of it as handing a stranger a copy of your house key. The lock cannot tell whether you sent them or they took the key without asking. Merchants and banks have no reliable way to separate helpers you approved from misuse, so many of them default to blocking.
The verification checks Muse met at Walmart and the blocks at Amazon are the visible result of that default. The bank statements add a question of responsibility. According to Barnum, the person whose credentials are used could be the one held liable.
In my view, what happens next depends on whether an approved route for agents emerges. Blanket blocking protects platforms and accounts. It also pushes agent builders toward password sharing and scraping, the very methods the banks call insecure. The eBay model points somewhere else, where a platform grants permission to specific agents, so access is explicit, limited and can be withdrawn.
Whether merchants, banks and agent developers agree on a standard for that kind of permission is the question these disputes have not yet answered.


