Amazon claims that Muse, which acts as a personal agent to perform multi-step tasks, does not identify itself while browsing and captures customer credentials to access account histories and place orders without Amazon’s consent.
This standoff highlights a growing conflict over "agentic shopping," where AI tools buy items on behalf of consumers, potentially disrupting the direct relationship between retailers and their customers.
For Amazon, the stakes include the protection of its $68 billion advertising business, which relies on users browsing its pages and viewing sponsored products.
The company argues that third-party applications should operate transparently and respect a merchant’s decision to participate, comparing the situation to the formal agreements held between food delivery apps and restaurants.
While Meta asserts that Muse uses secure storage to handle passwords without "seeing" them, Amazon maintains that the agent’s behavior violates its conditions of use.
This move follows similar efforts by Amazon to block shopping agents from Google, OpenAI, and Perplexity.
Currently, users attempting to shop via Muse on Amazon encounter a popup stating that unauthorized AI access is prohibited, though the companies remain in direct conversation regarding the issue.