Meta has paused its Model Compatibility Initiative (MCI) employee-tracking program following an internal data leak that exposed sensitive worker information. Company spokesperson Tracy Clayton confirmed the suspension while investigators assess the breach. The tool, rolled out to US employees in April, collected computer inputs including mouse movements, keystrokes, and screen content to train AI systems on human-computer interaction patterns.

Employees initially couldn't opt out of MCI, though Meta changed course after worker protests raised privacy and security concerns. The controversy reflects broader tensions around workplace surveillance and data collection practices in tech companies developing AI capabilities. A former employee described the security lapse as predictable, criticizing leadership for dismissing worker warnings about data safety risks.

The breach occurred when internal databases containing MCI-gathered information became accessible to all Meta employees, according to an internal security notice. Stephane Kasriel, Meta's AI research vice president, revealed the issue was discovered on June 18 and partially addressed within four hours, though additional access restrictions proved necessary. Some workers remained confused as the tool continued running on their devices.

Meta executives previously defended MCI as essential for training AI to mimic human computer behavior, positioning employees as ideal training subjects. However, the incident highlights how workplace monitoring initiatives can create security vulnerabilities. The company won't reactivate MCI until data protection controls are verified, and has collected enough data to evaluate the tool's long-term viability.