On Sunday night, users of Meta’s AI assistant Muse encountered an error message when attempting to purchase items on Amazon. The message, reported by GeekWire, stated that continued access by an unauthorized AI agent violated Amazon’s Conditions of Use, to which customers had agreed.

The blockage prevents Muse from completing transactions on Amazon’s platform, requiring users to switch to alternative shopping methods.

Amazon operates its own foundation models and maintains one of the most widely used inference platforms for e-commerce. The company has not indicated a legal obligation to permit third-party AI agents like Muse to interact with its shopping infrastructure, and may choose to restrict such access pending further development.

Industry analysts note that even with relatively low hallucination rates for AI systems, errors in order placement could lead to customer complaints and vendor disputes. Amazon may therefore delay integration of agent-driven commerce features until model reliability improves across additional release cycles.

The situation reflects broader tensions between major technology firms as AI capabilities expand into commercial applications.