An internal post seen by Refinitiv shows Meta Platforms is testing a “human
concierge” mode for its new AI personal assistant Muse, in which live agents
quietly handle some calls placed by the digital assistant. The company briefed
employees on the test last week after publicly enabling Muse’s calling feature.
Some staff raised privacy concerns, warning sensitive information could be
inadvertently disclosed to call‑centre contractors. A Meta spokesperson said
most employee feedback has been positive, the test is intended to gather user
input to refine safety and privacy measures before wider release, and the
company is working with merchants and will only launch the feature when ready
with appropriate disclosures.