TRENDING
An open Pacific oyster shell resting on a wooden chopping board
November 13, 2026
CNCF’s OpenShell-on-Kubernetes Guide Turns Agent Sandboxing Into a Two-Layer Split, but the Demo’s Seven Kyverno Policies Admit Secret Volumes, Oversized Init Containers and Unconfined Seccomp, and the OpenShell Release It Describes Is Five Behind the Current One
Close-up of the white locking bars and cam handles on the closed doors of a grey intermodal shipping container
November 13, 2026
Red Hat Published Eight Fixed Quay Releases Within Two Hours of Its CVSS 9.8 Signing-Key Advisory, but the CVE Page Still Lists No Fix, and the Sibling 9.9 Flaw’s Page Names Those Releases While None of Their Advisories Lists It
A white life ring hanging on a wooden post above an empty sandy beach and calm blue water, a rescue tool kept within reach before anyone needs it
November 13, 2026
How to Recover Lost Commits and Files in Git With reflog and fsck, and Snapshot Uncommitted Work Before a Script or AI Agent Can Wipe It
Three meerkats standing upright on a red sand mound at dusk, each looking in a slightly different direction
November 13, 2026
Cloudflare’s QuePaxa Post Turns Leader Elections Into a Hedging Delay, but in a Simulation of the Paper’s Protocol a Five-Millisecond Hedge Costs Every Slot Its Fast Path, and the 2020 Outage It Cites Lasted Six Hours Because Database Replicas Had to Be Rebuilt
A shredded-potato hash brown browning in a dark frying pan
November 13, 2026
Microsoft’s November Patch Tuesday Fixes 1,094 CVEs by Its Own Count, Not a “Record” 1,235, and the Only Exploited Bug Is an Outlook Flaw That Can Leak an NTLMv2 Hash Without the Usual Warning
13 Nov 2026
SXZ.io SXZ.io
  • Home
Search the Site
Popular Searches:
Technology Amazon AI
Recent Posts
A coal miner’s hard hat with an electric headlamp resting on a dark cloth
Attackers Put Miners on 21 N-central Servers Within Nine Hours of a Public Exploit, 19 of Them on the Hotfix 2 Build, and Apache Does Not Expect to Release a Fixed Axis 1 for the Flaw That Starts the Chain
November 13, 2026
Concrete dam wall with a footbridge along its crest holding back a calm reservoir below tree-covered hills, with a few people standing at the near end of the walkway
How to Shut Down a FastAPI Service Without Dropping Requests: Handle the Stop Signal, Drain Before You Close, and Survive a Rolling Restart Behind nginx
November 12, 2026
Close-up of an iridescent teal duck feather with a brown edge against a blurred green background
Apache Patches a Critical mod_http2 RCE Found Independently by Anthropic’s Scanner and Google Researchers, but 2.4.70 Shipped Nearly Five Days Earlier Without the Fix and the One-Line Patch Was Public for About 58 Hours Before the Advisory
November 12, 2026
SXZ.io SXZ.io
  • Home

Categories

Articles 296 Posts
News 302 Posts
Learning Hub 268 Posts
Home/News/Amazon Blocks Meta’s Muse AI Agent From Shopping, 12 Days After Its Launch
News

Amazon Blocks Meta’s Muse AI Agent From Shopping, 12 Days After Its Launch

Amazon blocked Meta's new Muse AI agent from shopping on its site, 10 days after a court denied its bid to revive a similar case against Perplexity.

September 21, 2026 5 Min Read
90

Amazon has blocked Meta’s new Muse personal AI agent from shopping on Amazon.com, 12 days after Meta launched it. It is the latest move in a monthslong campaign Amazon has run to keep outside AI shopping agents off its site, and this time it is leaning on a different legal argument than the one that already failed once this year.

Table Of Content

  • What Amazon says the problem is
  • A legal argument shaped by an earlier loss
  • Frenemies with a lot at stake
  • What Muse actually does
  • What happens next

As of Sunday night, people trying to use Muse to shop on Amazon started seeing a popup message: “Continued access by an unauthorized AI agent violates Amazon’s Conditions of Use, to which our customers have agreed,” according to GeekWire, which first reported the block. Amazon said it had tried, without success, to get Meta to voluntarily exclude Amazon from Muse’s shopping capabilities before cutting the agent off.

What Amazon says the problem is

Amazon’s complaint centers on three things: Meta never told Amazon that Muse would access its store, the agent doesn’t identify itself as it browses, and it appears to capture and store customer credentials, which Amazon says could create privacy and security risks.

“We think it’s fairly straightforward that third-party applications that offer to make purchases on behalf of customers from other businesses should operate openly and respect service provider decisions about whether or not to participate,” an Amazon spokesperson said in a statement reported by GeekWire. Amazon compared the arrangement to food delivery apps working with restaurants, or online travel agencies booking through airlines, relationships it says typically require the merchant’s agreement. “Agentic third-party applications such as Muse have the same obligations, and we’ve requested that Meta remove Amazon from the experience,” the spokesperson added.

Meta has previously said Muse “has no visibility into people’s passwords or payment methods,” and that any credentials a user shares “go into secure storage, so Muse can use them without seeing them, including passwords a person types into the browser themselves.” Meta did not immediately respond to GeekWire’s request for comment on the block itself.

The browsing behavior Amazon is objecting to appears to be a documented part of how Muse works. According to TechCrunch’s reporting on Muse’s launch, the agent runs inside its own “dedicated, secure computer with its own browser,” which Meta calls Muse Secure VM. When a service Muse needs offers a public API, it connects through that; when no API is available, Meta says Muse falls back to accessing the service through its own browser instead, the same way a human would, just without announcing itself as an automated agent. The Verge said it used Muse to successfully buy tank tops on Amazon during testing before the block took effect.

A legal argument shaped by an earlier loss

This isn’t Amazon’s first attempt to keep a third-party AI agent off its site, and the last one didn’t go well. Amazon won a preliminary injunction in March against Perplexity, blocking its Comet browser’s AI agent from shopping on Amazon under the Computer Fraud and Abuse Act. On August 4, the Ninth Circuit vacated that injunction, ruling that it was the user, not Perplexity, who was legally “accessing” Amazon’s computers when Comet shopped on their behalf, so Amazon was unlikely to win the underlying CFAA claim. The court denied Amazon’s request for a rehearing on September 10, just ten days before Amazon moved against Muse.

Per GeekWire’s reporting, the Ninth Circuit’s ruling left one door open for Amazon: claims built on contracts and terms of service, rather than federal computer-hacking law. That is exactly the route Amazon is taking with Muse. Instead of filing suit, it is pointing to its Conditions of Use and showing users a popup rather than going back to court, at least for now. Asked whether it would consider legal action against Meta, Amazon declined to comment, saying only that it is in direct conversation with the company. Readers who want the underlying case can read the Ninth Circuit’s opinion in Amazon.com Services, LLC v. Perplexity AI, Inc. directly.

Frenemies with a lot at stake

The standoff is unusual because Amazon and Meta are also business partners. Amazon products have been purchasable directly inside Facebook and Instagram since 2023, and Meta signed a multibillion-dollar deal in April to run agentic AI workloads on Amazon Web Services, according to GeekWire. Amazon’s own business stakes are also considerable: the company generated more than $68 billion in advertising revenue last year, a business that depends on shoppers browsing its pages directly rather than having an AI agent complete checkout on a page they never see.

GeekWire also reports that Muse is not the only agent Amazon has pushed back against this year: the company has spent the past year trying to keep outside shopping agents off its site generally, including moves against agents built by Google and OpenAI in addition to the Perplexity lawsuit.

What Muse actually does

Meta introduced Muse on September 8, describing it as a personal AI agent that can send emails, book travel, fill out forms, turn a recipe into a grocery list, and make purchases on a user’s behalf using Link by Stripe for checkout. It runs on Meta’s own Muse Spark model, which sxz.io covered earlier this month when Meta began offering steep API discounts in exchange for usage data. Muse is available on the web, iOS, Android, and inside WhatsApp, free to start, with paid Power ($20 a month) and Maximum ($100 a month) tiers for heavier use, per TechCrunch’s reporting.

The dispute is the latest example of a broader problem this site has covered repeatedly this year: platforms and AI agent makers disagree about who gets to authorize an agent to act on a user’s behalf, and on what terms. sxz.io recently covered Google’s own struggle with the same question when it opened Google Home to third-party AI agents, a case where the authorization problem ran in the opposite direction: Google inviting agents in, rather than Amazon trying to keep one out.

What happens next

For now, Amazon says it is talking directly with Meta rather than suing, and Meta has not publicly responded to the block. Whether that conversation produces a negotiated integration, similar to the food-delivery and travel-booking partnerships Amazon cited, or pushes Amazon back toward litigation this time built on contract law rather than the CFAA claim that failed against Perplexity, is the question both companies now have to answer.

Tags:

Agentic AIAgentic CommerceAI AgentsAmazonMeta

Share

A leaf insect camouflaged to look exactly like a leaf, with only its eyes revealing it is a living creature
Previous Post

How to Detect npm Packages That Hide Malware in Runtime Code, Not Install Scripts

An elaborate all-you-can-eat buffet counter with ornate brass fixtures and small bowls of food arranged along a glass display case
Next Post

Salesforce’s AI Agents Turn Per-Seat Licensing Into a Moving Target

No Comment! Be the first one.

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

Latest
13 Nov
Attackers Put Miners on 21 N-central Servers Within Nine Hours of a Public Exploit, 19 of Them on the Hotfix 2 Build, and Apache Does Not Expect to Release a Fixed Axis 1 for the Flaw That Starts the Chain
12 Nov
How to Shut Down a FastAPI Service Without Dropping Requests: Handle the Stop Signal, Drain Before You Close, and Survive a Rolling Restart Behind nginx
Trending
November 13, 2026
Attackers Put Miners on 21 N-central Servers Within Nine Hours of a Public Exploit, 19 of Them on the Hotfix 2 Build, and Apache Does Not Expect to Release a Fixed Axis 1 for the Flaw That Starts the Chain
November 12, 2026
How to Shut Down a FastAPI Service Without Dropping Requests: Handle the Stop Signal, Drain Before You Close, and Survive a Rolling Restart Behind nginx
November 12, 2026
Apache Patches a Critical mod_http2 RCE Found Independently by Anthropic’s Scanner and Google Researchers, but 2.4.70 Shipped Nearly Five Days Earlier Without the Fix and the One-Line Patch Was Public for About 58 Hours Before the Advisory
November 12, 2026
CNCF’s OpenShell-on-Kubernetes Guide Turns Agent Sandboxing Into a Two-Layer Split, but the Demo’s Seven Kyverno Policies Admit Secret Volumes, Oversized Init Containers and Unconfined Seccomp, and the OpenShell Release It Describes Is Five Behind the Current One
November 12, 2026
Red Hat Published Eight Fixed Quay Releases Within Two Hours of Its CVSS 9.8 Signing-Key Advisory, but the CVE Page Still Lists No Fix, and the Sibling 9.9 Flaw’s Page Names Those Releases While None of Their Advisories Lists It
November 11, 2026
How to Recover Lost Commits and Files in Git With reflog and fsck, and Snapshot Uncommitted Work Before a Script or AI Agent Can Wipe It

Related Posts

Rows of server racks in a data center representing network infrastructure targeted by botnets
News

C0XMO Botnet Shows Why Old Router Firmware Still Matters

June 7, 2026
Close-up of a USB flash drive, representing physical data-theft risk in office security incidents
News

Fake IT Support Is Now Walking Through the Front Door

June 7, 2026
A phone security app on a smartphone resting on a laptop keyboard.
News

Everest Forms Pro Flaw Is Being Exploited to Create Rogue WordPress Admins

June 7, 2026
A customer-support representative wearing a headset against a dark studio background.
Articles

The Meta AI Support Hack Was a Plain Old Authorization Failure

June 7, 2026
SXZ.io SXZ.io
  • [email protected]

Categories

Articles
Learning Hub
News

All Rights Reserved by SXZ.io ©2026