Meta AI Agent Muse Outsources Calls to Human Call Centers, Leaks Reveal

Meta’s latest conversational model, dubbed Muse, was marketed as a fully autonomous Meta AI agent capable of dialing customers without human help.

Meta AI agent’s hidden reliance on call‑center staff

Leaked internal memos show the system automatically transferred every outbound attempt to live operators, effectively turning the “AI‑only” promise into a human‑assisted service. “We route the call to a trained associate the moment the bot detects hesitation,” one engineer wrote, according to Reuters. The practice, described as a “fallback protocol,” was intended to keep conversion rates high; early tests recorded a ↑ 14% lift in successful connections versus pure AI dialing. Critics argue the approach blurs the line between genuine automation and deceptive marketing, especially as Meta touts the feature in investor briefings.

“Consumers deserve transparency about who is on the line,” a consumer‑rights group said.

The revelation arrives as the tech sector grapples with post‑pandemic hiring shortages, prompting firms to lean on hybrid solutions that combine bots with human labor. Analysts at Bloomberg predict that similar models could see a ↓ 8% drop in investor confidence if disclosure standards tighten.