Meta's ‘autonomous' AI Muse was quietly using humans to make phone calls

Meta's new AI assistant Muse promises to take care of everyday chores for you, from sending emails to booking trips but some users testing its latest feature may not have realised there was another human involved.

The Facebook parent company has been testing a “human concierge” system that allows contractors to quietly take over some phone calls made through Muse, according to internal company posts seen by Reuters.

Muse's calling feature is designed to do the routine phone work people often put off.

It can be used to book a haircut, check whether a shop has something in stock or get a quote from a contractor. The AI can call the business and speak on your behalf.

The problem came when some people on the other end refused to talk to an AI. Mark Zuckerberg-owned Meta came up with a solution and it was to have human contractors step in and complete some of those calls.

Employees warned that users could unknowingly share sensitive information with call-centre workers while believing they were dealing only with an AI system.

In one test, a contractor also made a racist remark during a call, according to Reuters.

Meta has now rolled back the human concierge feature.

A company spokesperson said the test was intended to gather feedback before any public release and that a calling feature would only be introduced when ready and with proper disclosures.

The controversy comes as Muse is enjoying a remarkable start.

Launched on September 8, 2026, the personal AI agent has already topped US app download charts, as Meta tries to challenge ChatGPT and Claude with an AI that doesn't just answer questions but actually does things for you.