Amazon has blocked Meta’s Muse AI agent from shopping on its marketplace, opening a larger debate about who controls the customer relationship when AI agents begin buying products on consumers’ behalf.
Meta launched Muse earlier this month as a personal AI agent that can browse websites, compare options, fill out forms, and complete purchases after receiving user approval. But when Muse attempted to shop through Amazon, the e-commerce company blocked its access.
Amazon says Meta had not received permission for Muse to operate on its marketplace and that the agent did not identify itself while browsing. Meta has not publicly responded to the dispute.
Behind the disagreement is a question that could become increasingly important for e-commerce: when consumers delegate shopping to an AI agent, does the relationship belong to the retailer, marketplace, or the company operating the agent?
Amazon’s Conditions of Use require AI agents to identify themselves in every request they make to its website.
The company argues that third-party services purchasing products on behalf of customers should operate transparently and respect merchants’ choice to participate. Amazon has raised similar concerns before, including in its legal dispute with Perplexity over the Comet browser’s shopping agent.
At the same time, Amazon is developing its own agentic shopping technology.
Its Buy for Me feature can purchase products from external brand websites when an item is not available directly through Amazon. Amazon argues that its approach is different because Buy for Me identifies itself and allows brands to opt out.
The disagreement is therefore not necessarily about whether shopping agents should exist. It is increasingly about the conditions under which they should be allowed to interact with other companies’ ecommerce systems.
The business implications go well beyond website access.
Amazon has built an ecosystem around consumers browsing, searching, comparing, and purchasing products inside its marketplace. That activity also supports its advertising business, which generated more than $68 billion last year.
An external shopping agent could change that journey considerably.
Instead of searching Amazon for a product and browsing several sponsored and organic listings, a shopper could simply tell Muse: “Find me the best wireless headphones under €200 with strong noise cancellation and good battery life.”
Muse could research the available options and potentially use Amazon only as the final place to purchase the selected product. In that scenario, Amazon still processes the sale, but Meta’s agent controls much of the discovery and decision-making process.
Meta may also have its own commercial interest in that position. CEO Mark Zuckerberg has said Muse could eventually take a small share of transactions it facilitates. Agentic commerce therefore introduces a new intermediary between retailers and their customers.
The Amazon-Meta dispute shows that agentic commerce will not develop without friction.
Marketplaces have commercial reasons to protect their customer relationships, advertising models, data, and shopping experiences. AI companies, meanwhile, want their agents to work across as much of the internet as possible.
Technical standards and commercial agreements may eventually determine how these systems interact. Amazon CEO Andy Jassy has already said the company is discussing commerce-agent integrations with other companies.
But brands and retailers cannot know which AI interface a future customer will use.
A shopper could begin with Muse, ChatGPT, Alexa, Gemini, or another agent that has yet to launch. The interface may change, while the underlying requirement remains remarkably similar: the agent needs accurate information about the products it is evaluating.
For e-commerce companies, this makes product data a more portable asset than any individual discovery channel.
The current disagreement between Amazon and Meta may be an early example of a much larger competition.
If AI agents become the primary interface through which consumers research and purchase products, companies controlling those agents could gain significant influence over which products shoppers encounter before reaching a marketplace or retailer.
Amazon, meanwhile, has strong reasons to keep discovery within an ecosystem it controls.
For brands and retailers, the outcome matters. But another practical lesson is that product discovery is becoming less tied to a particular storefront.
Whether a shopper starts on Amazon, Meta, Google, ChatGPT, or somewhere else, products need to be accurately represented wherever machines are comparing them.
The next e-commerce battle may therefore be about who owns the customer relationship. But for products to participate in that battle, AI agents first need reliable information to understand what they are buying.
Read further: News, AI, amazon, e-commerce, ecommerce, Icecat, Meta, product content
