Meta Exposed Data Internally From Its Controversial Employee-Tracking Program
An internal security notice reveals that Meta's employee laptop monitoring program, designed to train AI, inadvertently made sensitive data accessible to all employees, leading to an indefinite pause of the initiative.
Meta has encountered a significant security breach involving its employee laptop surveillance program. According to an internal security notice obtained by WIRED, data collected as part of the Model Capability Initiative—including keystrokes, mouse clicks, and screen content—was accessible to any company employee. Meta spokesperson Tracy Clayton initially confirmed the investigation but later stated the program is being paused indefinitely.
The security notice sent Monday indicated that employee data across 45,000 hive tables had been exposed, containing full prompts and transcriptions, private conversations, and performance data. Employees quickly voiced concerns in internal forums, stating the incident validated their previous objections to the workplace monitoring.
Meta's Chief Technology Officer Andrew Bosworth acknowledged that the program's implementation fell short of privacy standards. "Here we had misconfigured ACLs and we need to understand how that happened," he wrote in an internal post. Earlier, over 1,600 employees had signed a petition against the surveillance, warning of security and regulatory risks.
CEO Mark Zuckerberg had defended the program, claiming it was necessary for AI model training. However, after widespread protest, Meta began offering exemptions from monitoring. The incident is likely to worsen the morale crisis at Meta, where employees are already frustrated by layoffs and reorganization.
