Meta's laptop-surveillance program for AI training exposed sensitive employee data internally, reigniting concerns about privacy safeguards that staff had already protested.

What happened

Meta left employee data accessible to anyone inside the company as part of a program that began tracking US workers' corporate laptops in April. The exposed information included keystrokes, mouse clicks, screen content, full prompts, transcriptions, and private conversations from 45,000 employee data tables. An internal security notice was sent Monday; Meta's chief technology officer Andrew Bosworth said the program's implementation fell short of its privacy review standards.

Why it matters

More than 1,600 Meta employees had already signed an internal petition last month warning that the data collection posed security and regulatory risks, including potential breaches. The security failure validates those concerns and comes as the company faces ongoing morale challenges from mass layoffs, reorganizations, and a push to develop AI systems. Employees have described some new AI-focused roles as menial and demoralizing.

What to watch

Meta has begun offering exemptions to the monitoring—letting staff briefly turn off surveillance for sensitive tasks—but some employees are still demanding the tracking be stopped altogether. The incident marks a test of whether the company can rebuild trust after Bosworth apologized last week for 'atrocious' communication about the AI reorganization.