When Meta launched the Muse personal assistant on September 8, it promised to handle multi-step tasks ranging from calendar scheduling to online shopping. Within days, users reported a cryptic pop-up while trying to order through Amazon: “Continued access by an unauthorized AI agent violates Amazon’s Conditions of Use, to which our customers have agreed.” The message signals a decisive move by the e-commerce giant to bar the new AI from its marketplace.
The confrontation is more than a technical hiccup; it pits two of the world’s most powerful tech firms against each other in a battle over who controls the future of agentic commerce. While Amazon operates one of the largest inference platforms and owns a suite of foundation models, Meta has been aggressively expanding its AI capabilities, inviting users to let Muse shop, pay, and even negotiate discounts on their behalf.
How Muse attempts to shop on behalf of users
Muse functions as a conversational agent that can connect to external services through either public APIs or a built-in browser that mimics human interaction. Users supply credentials once, and Muse stores them in a secure virtual machine, claiming it never actually “sees” passwords or payment details. Before any purchase, the assistant asks for confirmation, and a secondary monitor called Sentinel must approve any outbound request. This architecture is marketed as a privacy safeguard, yet the ability to navigate Amazon’s site without identifying itself raises red flags for the retailer.
In practice, Muse can scrape product pages, fill carts, and click “Buy” much like a human would, but it does so programmatically. Early adopters have praised the speed at which Muse can locate discount codes, switch insurance policies, or assemble grocery orders. Its rapid ascent—reaching the top of the U.S. Apple App Store’s free list within a week—demonstrates strong consumer curiosity about AI-driven shopping assistants.
Amazon’s objections and legal posture
Amazon’s official stance is that third-party applications must operate transparently and respect the merchant’s right to opt out. The company argues that Muse never sought permission to access its storefront, does not disclose its presence while browsing, and could potentially capture and store user credentials—an activity that might breach both privacy regulations and Amazon’s own Conditions of Use. The retailer also warns that any erroneous order placed by an autonomous agent would leave Amazon to handle refunds, vendor disputes, and disgruntled customers.
Beyond the immediate technical concerns, Amazon has a history of defending its ecosystem against external bots. Earlier in the year, it secured a preliminary injunction against Perplexity’s Comet browser and updated its robots.txt file in late 2025 to block ChatGPT-style agents. The current blockage relies on contractual arguments rather than accusations of hacking, leveraging the platform’s terms of service to justify the denial.
Implications for the AI-shopping landscape
The standoff highlights a broader tension: who owns the consumer relationship when an AI acts as the purchaser? Visa and Mastercard have recently rolled out infrastructure to support AI-mediated payments, indicating industry momentum toward fully autonomous checkout experiences. Yet Amazon’s move suggests that large marketplaces may prefer to keep the transaction loop within their own AI tools, such as the “Buy for Me” feature launched in 2025, which explicitly identifies itself and offers brands an opt-out mechanism.
For Meta, the blockage is a setback, but the company maintains that Muse operates with an audit trail and never views raw passwords. The dispute may force both parties to negotiate an API-level partnership or a clearer consent framework, potentially shaping standards for future AI-shopping integrations. Meanwhile, developers of other agents—Google, OpenAI, and smaller startups—will be watching closely to see whether Amazon extends its policy to additional platforms.
Consumers stand to benefit from the competition once a sustainable model emerges. Transparent consent, robust security checks, and shared revenue arrangements could enable AI assistants to add genuine value without compromising the integrity of massive e-commerce ecosystems. Until then, the clash between Amazon and Meta serves as a reminder that the promise of seamless, hands-free shopping still faces real-world hurdles.
