Meta Tests Human Concierge for Muse AI Amid Privacy Backlash

Human Labor Behind the Muse AI Assistant

The Muse agent quickly climbed to the top of US app download charts, accumulating more than 2.5 million downloads according to data from Sensor Tower cited by Reuters. The digital assistant is designed to autonomously execute tasks such as sending emails, shopping, and booking travel. Its phone-calling feature enables users to instruct the agent to dial US businesses to handle practical errands, including booking haircuts, checking stock inventory, or obtaining contractor price quotes. Internal company communications revealed that Meta began utilizing human contractors to handle some of these phone calls behind the scenes. According to Reuters, internal testing indicated that involving human operators pushed the calling success rate to a range of 95% to 98%, significantly higher than the performance achieved by AI-only calling. The test mirrors an earlier experiment a decade prior when the company trialed a digital assistant named “M” within Messenger, which media reports estimated relied on humans for roughly 70% of its tasks.

Employee Backlash and Privacy Concerns

The disclosure of human-handled phone calls drew immediate criticism from internal staff who warned about potential data leaks. As reported by Reuters, one employee wrote in an internal post It’s baffling to me why we think this feature is worth the risk, adding that the system was one bug away from unnecessary information being leaked to human callers. Another worker shared a call transcript showing that a contractor made a racist reference while negotiating a bill with an internet and cable provider on the employee’s behalf. A vice president within Meta’s Superintelligence Labs unit responded to the backlash in internal posts, acknowledging that it was a miss to initiate the contractor testing without adequate disclosures. The executive confirmed that Meta had rolled back this feature temporarily while continuing to develop autonomous calling capabilities. She also issued an apology regarding the racist remark and stated that the specific human agent involved would not be assigned to any further Meta projects.

Company Response and Market Impact

Addressing the internal feedback, Meta spokesperson Daniel Roberts stated that employee responses were overwhelmingly positive regarding the broader goals of the test, emphasizing that the trial was intended to get feedback so we can implement safety and privacy protections and improve features before we release them publicly, according to Reuters. Roberts added, 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 rollout of Muse has coincided with a substantial rally in Meta shares, which have risen more than 20% since the product launch, adding over $200 billion to the company’s market capitalization and reaching a seven-month high. However, shares edged down 0.6% on Tuesday following the reports regarding the human concierge testing, as noted by Reuters.