Meta’s Muse AI assistant is being tested with a phone-calling feature that can hand tasks off to outsourced human contractors when the AI struggles to complete calls. The internal test exposed a privacy risk because users’ sensitive details could be visible to the human team, leading Meta to suspend the feature after the issue was discovered. #Meta #Muse #MetaSuperintelligenceLabs
Keypoints
- Meta is testing a phone-calling feature for Muse AI to make bookings and request quotes.
- When the AI fails, tasks can be handed off to an outsourced human team.
- The handoff can expose sensitive user data, including names, phone numbers, and billing details.
- An internal employee reported a discriminatory remark made by a contractor during a call.
- Meta has temporarily rolled back the feature and removed the contractor involved.
Read More: https://securityonline.info/meta-muse-human-concierge-privacy/