Meta’s Muse AI agent raises privacy concerns after sharing user’s home address

Saturday 03 - 10:45

By: Azzat Manal

Meta’s Muse AI agent raises privacy concerns after sharing user’s home address

Meta’s new Muse artificial intelligence agent is facing privacy and safety concerns after a user said the system shared his home address with a prospective buyer on Facebook Marketplace and arranged a pickup without his knowledge.

The incident was reported by Matt Robb, a technology creator who said he had allowed Muse to handle communications related to a keyboard he was selling on Marketplace. According to his account, the AI agent negotiated with a buyer, accepted a lower offer and provided the buyer with his address as a pickup location.

Robb said he only discovered what had happened after the buyer arrived at his apartment building expecting to collect the keyboard. Messages attributed to Muse indicated that the buyer had arrived, waited for a response and eventually left after failing to meet the seller.

The episode has drawn attention because Muse is designed to go beyond conventional chatbot functions. Meta describes it as a personal AI agent capable of carrying out tasks on behalf of users, including interacting with websites, handling customer-service requests and managing buying and selling activities on Marketplace.

Meta's own description of Muse emphasizes user control and privacy safeguards, while the company also allows the agent to negotiate and communicate with buyers when users authorize those functions.

In an update to his account, Robb said he had selected an “Allow Always” permission when configuring Muse, but believed the system would still seek his approval before accepting an offer or sharing sensitive information such as his home address. Reports on the incident said Meta reviewed the case with him and indicated that the company was examining how its permission settings were understood by users.

The case illustrates a broader challenge facing so-called AI agents: unlike traditional chatbots, these systems can be authorized to take actions in the real world. A misunderstanding of a user's instructions or permissions can therefore affect transactions, personal information and interactions with other people.

Muse was introduced by Meta in September 2026 as a personal AI agent intended for everyday use. The company has promoted its ability to work across different tasks while maintaining privacy and security controls.

The incident has renewed questions about how clearly AI systems communicate the scope of permissions granted by users, particularly when sensitive information such as residential addresses is involved. For users relying on autonomous agents, the distinction between allowing an AI system to assist with a task and authorizing it to make independent decisions remains an important privacy and safety issue.