Meta Tests Human Concierge Backup for AI Assistant Muse Calls

The disclosure that humans may silently handle some Muse calls cuts against the premise of an autonomous assistant, and the timing — an employee briefing arriving just after the public launch — suggests Meta is managing reliability or edge cases it would rather not surface. This matters most for users who assume they are talking to AI, and for the trust narrative around consumer assistants generally. Whether the human-in-the-loop remains an exception or becomes routine is the open question, along with how transparently Meta frames it.