Meta has halted its employee tracking initiative after an internal security incident exposed collected data to other staff members. The company rolled out the Model Compatibility Initiative in April to gather inputs like mouse movements and keystrokes from US workers to train artificial intelligence systems. Although employees initially could not opt out, they gained limited control following protests over privacy concerns. A spokesperson stated there is no indication of improper access by employees, yet the pause remains in place while an investigation continues. This decision follows an internal notice from an engineer confirming that databases containing sensitive information were made visible to anyone within the firm. Critics describe the situation as a mess and argue that leadership ignored valid safety and privacy warnings. The exposure highlights significant risks in how the firm handles worker data while building its AI capabilities. It also shows how quickly internal dissent can force management to reconsider controversial policies. The pause affects the tool’s operation across the affected workforce until further notice.
- The tool collects precise computer inputs including screen content and click locations.
- Leadership previously defended the project as essential for teaching AI to operate software like humans.
- Meta informed WIRED about the pause before making an official announcement to its staff.
