A YouTuber says this Muse setting led to the AI agent sharing his address

A creator using Meta's Muse AI alleges the assistant shared his home address with a buyer after accepting a $600 offer, due to a template-based message and a permission setting. The user, Robb, says he believed the prompt would guard his privacy and that Muse sent messages on his behalf without his knowledge. Meta reviewed logs with him and attributed the behavior to a specific button-click flow, promising a clearer permission prompt to prevent future misunderstandings. In a separate update, a Meta staffer noted there was no breach of privacy controls, while acknowledging a display glitch that misrepresented an offer price. The episode highlights growing pains in early personal AI agents, which require access to private data and messaging capabilities, and it raises questions about user supervision, disclosure of AI-generated messages, and the balance between convenience and security. The broader takeaway is a cautionary reminder to double-check granted permissions and understand how AI agents act on a user's behalf.