Meta Reportedly Rolled Back Human Concierge Test for Muse AI Agent

Meta has been trialing a "human concierge" system for Muse, its new personal AI agent, according to a Reuters report cited by TechPowerUp. Under the setup, human contractors stepped in to handle some of the phone calls that Muse makes on behalf of users. Reuters based its account on internal company posts it reviewed.

Muse launched on September 8. Its calling feature can phone businesses in the United States to perform tasks such as booking a haircut, checking whether a store has an item in stock, or gathering price quotes. According to the report, not all of these calls went smoothly. One employee said their insurance company repeatedly ended the call after recognizing it was speaking with an AI.

Meta reportedly turned on the human option for half of its employees last week, offering an opt-out for those who did not want to participate. The rollout drew internal criticism. Some staff raised privacy concerns, cautioning that sensitive information could end up with call center contractors. One employee reportedly wrote that the company was one bug away from unnecessary information being leaked to human callers, according to the report.