A group of Meta employees published an open letter demanding the company halt its Model Capability Initiative (MCI), which collects detailed computer-use data for AI model training. The program captures keystrokes, mouse movements, screen content, and navigation patterns from employee workstations. Workers raised concerns after leadership couldn't provide completed privacy reviews or clear mitigation strategies.

Employees learned about MCI through an internal announcement, but received no concrete answers about data protection measures. The petition highlights that executives have selective opt-out privileges while regular staff cannot avoid collection. This data inevitably includes sensitive information like Social Security numbers and health records, creating significant security risks.

The concerns stem partly from Meta's recent regulatory history. In September 2024, the Irish Data Protection Commission fined the company €91m for storing user passwords in plain text, resulting in four GDPR violations. More recently, a March 2026 incident saw an AI agent provide incorrect instructions that led to sensitive user data exposure.

The petition argues that collecting employee data without explicit consent contradicts Meta's stated mission of building AI responsibly. Signatories contend that any approach to AI training that exploits workers through non-consensual data extraction undermines the principles of empowering individuals. The letter demands explicit protection of employee computer-use data under CCPA and CPRA regulations.