Meta has been testing a "human concierge" for Muse -- human contractors quietly handling some of the phone calls placed through the AI agent. Reuters reported the test on September 22, citing internal company posts seen by its reporters.

The test followed the public launch of Muse's phone-calling feature. According to the report, some employees raised privacy concerns: calls handled by contractors could expose sensitive information unintentionally, with account details and internal plans potentially passing through call centers.

A Meta spokesperson, Daniel Roberts, said employee response had been "overwhelmingly positive" and described the purpose as gathering "feedback so we can implement safety and privacy protections and improve features before we release them publicly." On the calling feature itself: "We're working with merchants to continue improving this potential calling feature, and will only roll it out when it's ready and with the proper disclosures."

The disclosure question is the sharp one. A human fallback is not a scandal -- escalation to a person is often the responsible design. The issue is whether the business on the other end of the line knew the machine had handed off to a contractor, and what rules govern access, recording, retention, and oversight of those calls. Those answers are not in the report. An assistant can be useful without being fully independent; the trouble begins when users -- and the people they call -- are left to infer independence the system doesn't consistently possess.