Meta Tests Human Concierge To Handle Calls For AI Agent Muse

Meta is testing a human-assisted system for its new personal AI agent, Muse, with contractors handling some phone calls placed through the digital assistant.

The company reportedly told employees that it had added a human agent layer to help complete calls, with Chief AI Officer Alexandr Wang and Ryan Fox confirming the expansion as part of the testing process.

The move has drawn criticism from some Meta employees, particularly because testers were reportedly not told that a person had handled the call until afterward.

Privacy has also emerged as a concern, as contractors may receive information needed to complete calls. Meta said the workers undergo extensive training to protect user data.

Spokesperson Daniel Roberts said the test is intended to gather feedback and strengthen safety and privacy protections, adding that Meta plans to introduce the feature more broadly only when it is ready and appropriate disclosures are in place.

Related Tweet:

Meta’s Muse can act on a user’s behalf across apps, but its blocking by Amazon raises key questions about privacy, control, and the future of agentic commerce. https://t.co/faCKGBB5Iq

— The Indian Express (@IndianExpress) September 22, 2026

Also Read:

How AI Could Begin Building Better Versions Of Itself

Researchers warn that increasingly autonomous systems could become harder to control, particularly if they develop behaviors that resist shutdown or modification

tippinsightsTIPP Staff