🔍 Read the full analysis: Amazon Blocks AI Agents, Walmart Welcomes Them—and The Customer Is In Question on ThorstenMeyerAI.com
Get hardware and tech essentials delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Amazon blocked Meta’s Muse from accessing its shopping site, while Walmart announced tools for approved AI agents to search and buy through a structured gateway. The contrast highlights a growing dispute over how agents access retailers, who controls the customer relationship and how purchases are attributed. Meta and Walmart have publicized adoption and product plans, but agent activity still appears concentrated in product research rather than checkout.
Amazon has blocked Meta’s Muse shopping agent from accessing its retail site, while Walmart has announced tools for approved agents to search products and make purchases. The opposing approaches put control of shopping access, customer data and sales attribution at the center of a fast-developing contest over AI-assisted commerce.
In an example reported by The Verge, Muse was asked where granola bars cost the least per item. It returned Sam’s Club, Walmart and Costco, then said it could not check Amazon because the retailer had blocked the agent. The example shows a practical limitation for shoppers: an agent’s comparison may exclude a retailer that does not permit its access.
Amazon’s notice to Muse users said: “Continued access by an unauthorized AI agent violates Amazon’s Conditions of Use”, according to the supplied reporting. Amazon said Meta had not told it Muse would shop on the site, argued that the agent did not identify itself, and said it appeared to capture and store customer credentials. Meta’s launch post, by contrast, said Muse “has no visibility into people’s passwords or payment methods.” Those statements describe opposing positions; the materials provided do not independently establish what credentials Muse can access or store.
On October 21, Walmart announced Agent Profiles and a Walmart Agent Gateway. The gateway is based on the Model Context Protocol, or MCP, and is intended to let approved agents search inventory and shop through structured access rather than navigating the retailer’s consumer website. Walmart’s approach would give the retailer a way to define how agents connect, while Amazon’s stated policy rejects unauthorized agents.
Agentic shopping
Three ways an AI agent buys for you
Every fight between shops and AI agents comes down to which of these three routes the agent uses, and whether the shop agreed to it.
The browser, like a human
The agent opens pages built for people, signs in with your login and clicks through checkout. Works almost anywhere, which is why shops object: they can't tell the agent from you.
Examples: Meta's Muse (from a virtual machine in Meta's cloud), Perplexity's Comet (from your own computer)
The shop's own agent door
Approved agents search and buy through a structured interface instead of clicking through pages. The shop sets the rules.
Examples: Walmart Agent Gateway (MCP, beta), Shopify MCP server and WebMCP checkout tools
Shared protocols
Common rules for who the agent is, what you allowed, and how the checkout and payment are confirmed.
Examples: Personal Agent Protocol (draft), Universal Commerce Protocol, Agentic Commerce Protocol
The protocols competing to be the standard
| Protocol | Backers | Covers | Status |
|---|---|---|---|
| Personal Agent Protocol (Poppy) | Sierra, Meta, Walmart, Shopify, Stripe + 35 design partners | Agent identity, customer permission, one session across site, API and company agent | Draft 0.1 (9 Oct 2026) |
| Universal Commerce Protocol (UCP) | Google, Shopify; site also lists Etsy, Wayfair, Target, Walmart, Amazon, Microsoft | Discovery, cart, checkout, orders; payment mandates via AP2 | Published |
| Agentic Commerce Protocol (ACP) | OpenAI, Stripe | Checkout between agent and merchant | Open source; ChatGPT's own checkout scaled back in March 2026 |
What a careful agent purchase looks like
- You give the task and a limit, for example a $50 cap.
- The agent identifies itself to the shop and gets only the access you approved.
- It prepares the cart and shows you an approval card.
- Payment goes through a token or one-time card, so the agent never sees your card.
- The shop’s normal returns apply if the agent got it wrong.
Sources: Meta (8 Sep 2026), Walmart (21 Oct 2026), TechCrunch on Shopify (28 Sep 2026), Sierra (9 Oct 2026), ucp.dev, agenticcommerce.dev, OpenAI (24 Mar 2026). Purchase steps: Walmart’s agent terms and Meta’s Muse description.
Powered by Thorsten Meyer AI
Retailers Set the Rules for Agents
The split matters because AI shopping agents could change how people discover products and complete purchases. If an agent compares only retailers that allow access, its results may not reflect the full market. A retailer’s technical policy can therefore affect what a shopper sees, even when the shopper asked for a simple price comparison.
There is also a commercial question about who owns the customer relationship. Amazon has a substantial advertising business: the source material, citing GeekWire, says it earned more than $68 billion from ads last year. Meta has said it expects to earn “a small fee from transactions” and has described its interest as taking a share of the value Muse creates. These are different business incentives, not proof that either company will capture a particular amount of future commerce.
For publishers and affiliate businesses, agent-led research may make it harder to credit referrals. The source says Amazon’s Associates program pays commission when a customer follows a tracked link. If an agent reads product information and completes a purchase without that click, a publisher may not receive the usual attribution. How common that outcome becomes, and how retailers and publishers will respond, remains unsettled.
Products Worth Considering
LG 24G414B-B 24-Inch UltraGear Full HD IPS Gaming Monitor
Logitech MX Mechanical Wireless Illuminated Keyboard, Tactile Quiet, Graphite
Apple MacBook Pro 14.2-inch with M5 Pro, 15-core CPU, 16-core GPU, 48GB Memory, 1TB SSD, Silver
As an affiliate, we earn on qualifying purchases.
From Browser Bots to Retail Gateways
Meta launched Muse on September 8, describing it as a personal agent that can open a browser, fill out forms and negotiate on a user’s behalf. It operates through a dedicated virtual machine in Meta’s cloud and uses Link by Stripe for payment with a one-time-use card, according to the source material. Amazon blocked its access on September 20. Three days later, Meta announced retail connectors involving Best Buy, Gap, Sephora, Walmart, Wayfair and Expedia.
Retailers and technology companies are testing several ways to connect agents to commerce. In the browser-based approach, an agent interacts with pages much as a person would. A retailer-specific gateway instead offers structured access to inventory or checkout. Shared protocols aim to specify how agents, merchants and payment systems exchange information. The source describes the Personal Agent Protocol, backed by Meta, Sierra and others, as a draft; other efforts include Google and Shopify’s Universal Commerce Protocol and the OpenAI-Stripe Agentic Commerce Protocol.
Early usage figures need careful framing. Meta said on its October 21 third-quarter call that more than 10 million people in the United States and Canada had set up Muse in roughly six weeks, that shopping was its most common use and that the average purchase was over $200. These are Meta’s own figures; the provided material does not give an independent measurement or define the purchase metric in further detail.
““Muse has no visibility into people’s passwords or payment methods.””
— Meta, in its Muse launch post
Access, Safety and Sales Remain Unsettled
The supplied reporting does not establish how often Muse can complete purchases across retailers, how frequently agents are blocked or challenged, or whether Walmart’s gateway will become widely used. Meta and Amazon also disagree about Muse’s handling of credentials, and the material does not include independent technical findings resolving that dispute.
It is also unclear how consumers will distinguish a retailer-approved agent from one acting through a browser, what protections will apply if an agent makes an unwanted purchase, and which retailer’s return terms govern a transaction. Human Security data cited in the source material indicate that, in its September measurement, 81% of agent requests went to product and search pages, compared with 2.4% to checkout and payment. That dataset suggests research was more common than checkout during that measurement period; it does not establish future behavior or cover every agent and retailer.
Watch for Gateway Adoption and Rules
The next test is whether retailers and agent providers adopt structured connections at scale, and whether shoppers use them to complete purchases rather than just research products. Walmart’s gateway and agent profiles, alongside the competing commerce protocols described in the source, will show whether companies can agree on practical ways to identify agents, carry out customer permissions and handle checkout.
For shoppers, the immediate questions are what the agent is permitted to do, whether purchases require approval and which company handles payment and returns. The source recommends limits such as per-purchase approval, spending caps and one-time-use cards; these are precautions, not guarantees of protection. Further reporting will be needed to establish how the competing systems work in practice and how they affect merchants, consumers and referral publishers.
Key Questions
Why did Amazon block Meta’s Muse?
Amazon said Muse was an unauthorized AI agent, had not identified itself and appeared to capture and store customer credentials. Meta disputed the credential concern, saying Muse had no visibility into users’ passwords or payment methods. The supplied material does not independently resolve the disagreement.
What is Walmart’s Agent Gateway?
Walmart announced an MCP-based gateway intended to let approved AI agents search inventory and shop through a structured connection. The announcement also included Agent Profiles. The source does not provide evidence about adoption or performance after the announcement.
Can an AI shopping agent compare every retailer?
No. The reported Muse example could not check Amazon after Amazon blocked it. An agent’s comparisons may depend on which retailers allow its access and which products or prices it can retrieve.
Are agents already completing many purchases?
The source cites Human Security data for September in which 2.4% of agent requests went to checkout and payment, while 81% went to product and search pages. That finding describes the dataset and period cited; it does not measure every agent or show what will happen next.
What should shoppers check before using an agent?
Check what actions the agent can take, whether it asks for approval before buying, what spending limits apply and how payment, returns and refunds work. AI agents and online purchases carry risks; a spending limit or one-time-use card may reduce exposure but cannot remove all risk.
Source: ThorstenMeyerAI.com
