Meta Muse Is People
Spyglass ● Covered by 19 sources
Meta tested a “human concierge” for its Muse AI, with contractors secretly helping with phone calls. It undercuts the idea of a fully autonomous assistant, and privacy worries are already inside Meta.
Based on reporting by Spyglass — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Meta has been testing a human-assisted version of Muse, its new personal AI assistant, by having contractors quietly handle some of the phone calls the agent places. The test was described in internal posts seen by Reuters, and employees were told about it last week, after Meta had already publicly launched Muse’s phone-calling feature.
Muse itself is meant to do more than chat. Meta says the assistant can carry out tasks like sending emails, shopping, and booking travel. The phone feature is supposed to let users tell Muse to call US businesses and deal with routine errands such as booking haircuts, checking whether a store has an item in stock, or asking contractors for quotes.
But the whole setup brings back an old trick from the earlier days of AI: put a human in the loop when the software can’t quite pull it off alone. Meta used a similar hybrid approach with “M,” its long-ago virtual assistant from the Facebook era. The difference now is that users are supposed to believe the AI is finally good enough to work on its own.
That belief is doing a lot of heavy lifting. Muse has topped US app download charts in the two weeks since it launched, but some Meta employees raised privacy concerns about people in call centers hearing sensitive information. One wrote that the feature felt risky and warned that a bug could leak unnecessary information to human callers.
Meta appears to have been moving fast to keep pace with rivals, and the company seems to have rolled the feature back for now. The broader pattern is obvious enough: the AI may be smarter, but the edges of the product still need people, at least when the job is making a phone call.
My take — AI-written commentary, not fact-checked reporting
This is the part of AI everyone pretends not to notice: the shiny agent often turns into a call center with nicer branding. Meta can call it beta all it wants, but if the pitch is autonomy and the delivery is contractors, users are right to feel mugged by the fine print. The industry keeps selling robots and shipping partial humans.
Read more about this at: Spyglass
