Meta rolled back its Muse human-assisted calling test after a contractor made a racist remark during an employee’s bill-negotiation call. The test involved roughly half of Meta’s employees, who were not clearly told a human might take over.
Meta apologized to the employee and said the contractor would not work on any of its projects again. A Meta Superintelligence Labs vice president called the lack of disclosure “a miss.” Spokesperson Daniel Roberts said employee feedback was “overwhelmingly positive” and described the test as groundwork for safety and privacy protections before any public release. Meta’s stated position, according to Reuters, is that the feature may return once protections are in place.
Reuters reported that human-handled calls succeeded about 95–98% of the time, compared with substantially lower rates when Muse handled calls on its own. The feature was designed for routine tasks such as booking appointments and checking product availability; businesses were reportedly reluctant to engage with AI callers.
