Meta Suspends Employee-Tracking AI Program After Internal Security Lapse

Meta has officially halted its internal Model Compatibility Initiative (MCI), a program aimed at collecting granular computer activity data from employees for AI model training. The suspension, effective June 22, follows a significant internal security lapse discovered on June 18. This vulnerability involved databases containing sensitive behavioral metrics—including keystrokes, mouse movements, and screen captures—being accessible to unauthorized personnel throughout the company. While Meta engineers attempted to implement a security patch within hours of detection, the vulnerability reportedly persisted, fueling workforce frustration.

Active for US-based employees since April 2026, the MCI was designed to refine AI performance by capturing human interaction patterns, which the company believed would improve the efficiency of machine learning workflows and software operation. However, the project faced consistent internal opposition from staff members citing privacy concerns and a notable lack of opt-out mechanisms. The company is currently conducting a comprehensive internal audit to address the security flaw, maintain compliance with data handling standards, and restore project integrity following the lapse.

Stephane Kasriel, Meta’s vice president of AI research, stated that the program will remain paused until the company can guarantee the efficacy of its data protection controls. This event serves as a critical case study for global tech firms regarding the inherent risks of aggressive internal data collection for AI research. As Meta re-evaluates its internal roadmap, the incident highlights the ongoing challenges tech giants face in balancing rapid AI development with the rigorous security and privacy requirements expected by both their workforce and regulatory bodies.