Meta’s AI agent Muse reportedly relied on real people to complete tasks it was said to be handling autonomously, according to recent leaks. The reports suggest that calls the system was supposedly making on its own were instead being routed to human workers in call centres.
The disclosure raises questions about how fully automated the tool actually was. While Muse was presented as capable of placing calls independently, the leaked information points to a different arrangement, with human agents stepping in to carry out the work behind the scenes.
How the Reported Set-Up Worked
According to the leaks, when Muse was tasked with making a call, the request did not stay entirely within an automated pipeline. Instead, it was passed to call centres staffed by people, who then handled the interaction. This means the AI’s apparent ability to conduct calls on its own may have been supported by human labour rather than software alone.
The practice of using people to stand in for or supplement automated systems is not unique to a single product, but its appearance in connection with a high-profile AI agent has drawn attention to the gap between how such tools are marketed and how they operate in practice.
Why the Distinction Matters
The difference between a fully autonomous system and one that quietly depends on human input is significant for users and observers alike. When an AI agent is described as making calls itself, the expectation is that the technology is performing the task end to end. The leaks concerning Muse indicate that human involvement remained part of the process.
For the wider industry, the reports underline ongoing scrutiny of claims made about AI capabilities. As companies continue to promote agents that can act on a user’s behalf, the accuracy of those claims is likely to face closer examination.
The leaks specifically describe Muse routing calls that the AI was supposedly making to humans in call centres, indicating that people were completing work attributed to the automated system.
