Meta's Muse agent has topped the app download charts in the US for a couple of weeks, with over 2.5 million downloads, and the autonomous AI agent can act like a personal assistant by sending emails, booking travel, shopping, and organizing calendars and meetings. One of Muse's features is phone calling, which lets the agent dial phone numbers in the US and talk to people on the other end to handle things like making bookings or checking whether a store has a particular item in stock.

According to a new report over at Reuters, Meta is reportedly testing outsourcing these AI agent calls to a real "human concierge." Yes, this would mean a human would make a call on behalf of a Muse AI agent, to increase the success rate based on the "lower percentage of AI calling."
This is one of the first reports of an AI agent's work being outsourced to a human, highlighting the reliance on human labor for autonomous AI.
Citing internal communications and sources within Meta, the report adds that employee feedback on the move to outsource AI agent phone calls to humans raises privacy concerns, especially since it could involve sharing sensitive personal information with call center workers contracted by Meta to handle Muse's phone calls. "We are one bug away from unnecessary information being leaked to human callers," an unnamed Meta employee reportedly said.

Frequently Asked Questions
Open a question for an answer from TweakTown's coverage of this news, or ask your own below.
How does Muse’s phone-calling feature work for bookings and store inventory checks?
What kinds of personal information could be shared with the human concierge handling Muse calls?
Why would Meta use a human caller instead of Muse’s AI for phone calls?
How did the outsourced ISP support call lead Meta to pause the plan?
Have a question that isn't listed here? Ask below, and TweakBot will answer it.
It got messier when a human caller acted inappropriately. Apparently, one outsourced phone call to handle an internet service provider issue led to the "human agent" making a racist remark, prompting Meta to put a pin in the plan. Apparently, the caller in question will not be involved in any Meta project moving forward, and the company plans to "continue improving this potential calling feature, and will only roll it out when it's ready and with the proper disclosures."







