XAI Faces Employee Backlash Over Facial Data Collection for AI Training

Elon Musk's AI startup xAI launched an internal initiative called "Project Skippy" to train its chatbot Grok in recognizing and interpreting human facial expressions and emotions by having employees record videos of themselves. Over 200 employees, including AI tutors, participated in 15- to 30-minute conversations playing both user and AI assistant roles, but many expressed discomfort and concerns over how their facial data and likeness might be used, with some opting out. The consent form granted xAI perpetual rights to use participants' likenesses for training and potential commercial applications, exacerbating privacy worries despite assurances that the videos would remain internal and not be used to create digital replicas. Employees were guided to have natural yet sometimes personal or provocative conversations during recordings, adding to unease amid broader concerns about the ethical handling of personal data. The project coincided with controversies surrounding Grok's behavior and xAI's development of avatars, such as anime companions with provocative features, intensifying employee skepticism about data usage and corporate safety culture. This episode reflects wider industry debates on AI ethics, privacy, and the boundaries of employee consent in AI training practices.