Amazon has blocked Meta’s new Muse AI agent from shopping on Amazon.com, highlighting a growing battle over how AI agents interact with online platforms and who controls the digital shopping experience.
The relationship between consumers and the internet could be entering a new phase.
For years, people have visited websites and apps themselves, searched for products, compared prices, logged into accounts and completed purchases. Artificial intelligence is beginning to change that model. Instead of navigating the internet manually, users can increasingly ask an AI agent to carry out tasks on their behalf.
That shift is now creating tension between some of the world’s largest technology companies.
Amazon has blocked Meta’s Muse, a newly launched personal AI agent, from accessing its retail platform to shop on behalf of users. Amazon says Meta did not obtain authorization for Muse to access its store and has raised concerns about how the agent interacts with its website and customer information. Meta, meanwhile, says Muse is designed with security controls that prevent the agent from seeing users’ passwords and payment information. (GeekWire)
The dispute could become an important early example of a much larger question: Who controls the internet when AI agents start acting on behalf of users?
What Is Meta Muse?
Meta introduced Muse on September 8, 2026, describing it as a personal AI agent designed to do more than answer questions.
Traditional AI chatbots generally respond to prompts. An agent such as Muse is designed to take actions.
Meta says Muse can help users with tasks such as sending emails, booking travel, shopping and organizing longer-term projects. It operates through a dedicated virtual computer called Muse Secure VM, which includes its own browser and stores the information and credentials needed to interact with connected services. (About Facebook)
The idea is relatively simple from a user’s perspective.
Instead of searching Amazon for a product, comparing listings and adding an item to the cart, a user could tell Muse what they want and allow the agent to handle the process.
That represents a fundamental change in how people could interact with online services.
The website is no longer necessarily the primary interface. The AI agent becomes the intermediary.
Why Did Amazon Block Muse?
Amazon says the issue is not simply that an AI agent is visiting its website.
According to Amazon, Meta did not notify the company that Muse would access Amazon’s store, and the agent did not identify itself while browsing. Amazon also raised concerns about privacy and security surrounding customer credentials and account information. (GeekWire)
Users attempting to shop on Amazon through Muse began seeing a message stating that continued access by an unauthorized AI agent violated Amazon’s Conditions of Use. (TechCrunch)
Amazon says it had asked Meta to remove Amazon from Muse’s shopping experience before taking the step of blocking the agent.
The disagreement therefore comes down partly to authorization.
Amazon operates the marketplace, controls access to its systems and has its own rules governing automated activity. Meta is creating an AI agent intended to act on behalf of consumers across the internet.
The two companies have different interests in how that interaction should work.
Meta Says Muse Protects Sensitive Information
Meta has presented Muse as a system built around security and user control.
According to the company, Muse runs inside a dedicated secure virtual machine. A separate system called Sentinel controls whether Muse can access the internet, and Meta says Sentinel must approve actions before they are carried out. (About Facebook)
Meta also says Muse cannot see users’ passwords or payment methods.
Instead, credentials supplied by users are stored securely so Muse can use them without directly viewing them. Meta says the system also asks for user approval before sensitive actions such as sending an email or making a purchase. (About Facebook)
These protections are central to Meta’s argument that Muse can operate safely as a personal agent.
Amazon, however, is raising a different question: whether Meta’s system should be permitted to access Amazon’s platform in the first place.
That distinction is important.
An AI company can build security protections into its own product, but websites and platforms still control how third-party software interacts with their services.
The Bigger Battle Is Over AI Shopping
The Amazon-Meta dispute is part of a much larger transition taking place in online commerce.
Amazon itself has invested heavily in AI-powered shopping.
The company combined its Rufus shopping assistant and Alexa+ capabilities into Alexa for Shopping, an agentic AI system designed to help customers research products, compare options, monitor prices and automate certain shopping tasks. (Amazon News)
Amazon says its shopping AI has already become a significant part of its retail strategy. In its second-quarter 2026 update, the company said more than 350 million customers had used Alexa for Shopping over the preceding 12 months, while active users nearly doubled and interactions increased more than fivefold year over year. (Amazon News)
That means Amazon isn’t opposed to AI-powered shopping itself.
Instead, the current dispute involves who gets to operate the shopping agent.
Amazon is developing its own AI shopping systems. Meta is developing a separate agent that can interact with Amazon’s marketplace.
That creates a potential conflict over the role of the retailer, the role of the AI provider and ultimately the relationship with the customer.
AI Agents Could Change How People Shop
Consider how online shopping works today.
A customer might:
Search Google.
Visit several websites.
Read reviews.
Compare prices.
Open Amazon.
Search for a product.
Read product information.
Add an item to a cart.
Enter payment details.
Complete the purchase.
An AI agent could potentially compress much of that process into a conversation.
A user could say:
“Find me a lightweight laptop under $1,000 with at least 16GB of RAM and good battery life.”
The agent could potentially research products, compare specifications and prices, identify suitable options and eventually place an order after receiving approval.
That sounds convenient for consumers.
But it creates a major problem for websites.
If the customer no longer spends significant time browsing the retailer’s website, the retailer could lose control over a large part of the customer journey.
Why This Matters to Amazon’s Advertising Business
Amazon isn’t just an online store.
Its website is also a massive advertising platform.
Sponsored products, recommendations and other advertising systems depend heavily on customers browsing Amazon’s pages and interacting with its search and product listings.
An external AI agent could change that.
Instead of showing a customer dozens of search results, an AI agent might select three products and present the options directly to the user.
The consumer gets a simpler experience, but the retailer could lose some of the opportunities to influence what the customer sees.
This is one reason agentic shopping could become much more significant than simply another feature in an AI chatbot.
It could alter the economics of online commerce.
Meta Is Taking a Different Approach
Meta has been building Muse around the idea that AI should connect to many of the services people already use.
At Meta Connect, the company announced additional partnerships and integrations involving shopping, payments, productivity tools and other services. Meta has also opened its platform to developers that want to build connectors for Muse. (TechCrunch)
That means Meta’s ambition goes beyond creating another shopping assistant.
It is attempting to create an AI intermediary that can interact with numerous online services.
If that model succeeds, consumers might increasingly interact with companies through their AI assistant rather than directly through individual websites.
For example, a person might ask Muse to:
Find a product
Compare prices
Book a hotel
Make a restaurant reservation
Send an email
Schedule an appointment
Order groceries
Track an order
The AI becomes the layer connecting the consumer to different services.
Shopify Is Taking a Different Approach
The response from other companies illustrates that there is no single industry position on AI agents.
While Amazon has blocked Muse, Shopify has moved toward integrating with Meta’s system.
Meta has announced partnerships that allow Muse to interact with Shopify merchants and use supported payment and shopping infrastructure. TechCrunch reported that Meta has also announced integrations involving companies such as Best Buy, Gap, Sephora, Walmart and Wayfair. (TechCrunch)
The contrast is significant.
Amazon is effectively saying that external AI agents need permission to operate on its platform.
Shopify is building infrastructure that allows merchants to participate in agent-driven commerce.
The difference could become an important competitive factor as AI shopping develops.
This Isn’t Amazon’s First Conflict With AI Shopping Agents
Amazon’s decision regarding Muse is also part of a broader pattern.
The company has previously taken action against other AI-powered shopping tools.
Amazon sued AI company Perplexity over its Comet browser and its ability to access Amazon for shopping-related tasks. The dispute raised similar questions about whether an AI agent should be able to navigate an online marketplace on behalf of a consumer without the marketplace’s explicit participation. (GeekWire)
The underlying issue is becoming increasingly clear.
AI agents can behave differently from traditional search engines.
A search engine sends users toward websites.
An AI agent can potentially enter the website, perform actions, make decisions and complete transactions.
That makes the relationship between the agent and the website much more complicated.
Who Owns the Customer Relationship?
One of the biggest questions emerging from agentic commerce is deceptively simple:
Who is the customer interacting with?
Suppose someone asks Muse to purchase a pair of headphones.
The customer might never visit Amazon.
Muse identifies the product, accesses Amazon, selects the listing and facilitates the purchase.
From the customer’s perspective, Muse was the interface.
From Amazon’s perspective, however, the transaction occurred on Amazon’s platform.
That creates competing interests.
Retailers want to maintain relationships with customers, control product presentation and protect their data.
AI companies want their agents to work across as many services as possible.
Consumers generally want convenience.
Those three objectives will not always align.
Security Is Another Major Issue
The technology also introduces new security challenges.
An AI agent that can browse websites and perform actions needs access to accounts, credentials and potentially sensitive personal information.
Meta says Muse was specifically designed around these concerns, including isolated computing environments, secure credential storage and approval mechanisms. (About Facebook)
But AI agents also create new attack surfaces.
If an agent can read email, browse websites, make purchases and access accounts, a malicious website or compromised service could potentially attempt to manipulate the agent.
That makes the security architecture surrounding AI agents extremely important.
The industry will likely need standards for identifying agents, authenticating them, controlling permissions and making their actions transparent to websites and users.
The Rise of Agent-Friendly Websites
The dispute could eventually encourage websites to build explicit systems for AI agents.
Instead of allowing an agent to behave like a human user through a browser, a company could provide an official API or AI connector.
The website could then define exactly what the AI is allowed to do.
For example:
Search products
Compare prices
Add items to a cart
Check availability
Place orders
Process payments
Each action could have its own permission level.
That would give companies more control while still allowing consumers to use AI assistants.
Meta’s growing connector ecosystem is moving in this direction, while Amazon’s response suggests that companies may also insist on controlling whether and how their services participate. (TechCrunch)
What Happens Next?
The immediate question is whether Amazon and Meta can reach an agreement that allows Muse to interact with Amazon under conditions acceptable to both companies.
Amazon has said it is in direct discussions with Meta, while declining to comment on whether legal action could follow. (GeekWire)
The longer-term question is considerably bigger.
If AI agents become a common way of navigating the internet, websites may need to rethink their entire architecture.
The web was originally designed around humans clicking links and filling out forms.
Search engines later changed how people discovered information.
Social networks changed how information was distributed.
AI agents could change how people act on the internet.
Instead of visiting ten websites, a user might give one instruction to an AI system and let the agent coordinate everything.
The Next Battle for Big Tech
Amazon’s decision to block Meta’s Muse may therefore be more than a disagreement between two technology companies.
It is an early sign of a broader battle over the future of online commerce.
Meta wants Muse to become a personal AI agent capable of carrying out tasks across the internet. Amazon is building its own AI shopping systems while maintaining control over its marketplace. Shopify is experimenting with ways for merchants to participate in agent-driven commerce. Other companies are likely to make their own choices as the technology develops. (About Facebook)
For consumers, the potential benefit is straightforward: less time spent navigating websites and more tasks handled automatically.
For technology companies, however, the stakes are much higher.
The companies that control AI assistants could increasingly influence how people discover products, choose services and complete transactions online.
That means the next major technology platform may not be a website or an app.
It could be the AI agent sitting between the user and the internet.
And Amazon’s decision to block Meta’s Muse shows that the competition over who controls that new layer has already begun.