Meta Transforms Employee Keystrokes into Training Data for Autonomous Artificial Intelligence
Menlo Park, Wednesday, 22 April 2026.
Meta is now recording employee keystrokes and screen clicks to train autonomous AI agents, leveraging internal workforce data to build technology that could ultimately automate those very jobs.
The Privacy Paradox in Corporate Data Collection
The rollout of the MCI tool has sparked internal controversy among Meta’s staff, who are effectively being tasked with training the very systems designed to automate their roles [2][3]. Management has pitched the surveillance as a passive opportunity for employees to “help our models get better simply by doing their daily work,” assuring staff that the collected data will not be utilized for performance assessments [1][2]. However, employees face a significant surrender of workplace privacy without receiving any corresponding increase in compensation for their dual role as workers and AI trainers [2].
Workforce Restructuring and Global Friction
The push for autonomous AI agents arrives against a backdrop of severe workforce contractions. Meta is preparing to lay off 10 percent of its global workforce starting on May 20, 2026, with executives reportedly considering additional cuts [1][2]. This downsizing trend is mirrored across the broader tech sector, with companies like Amazon and Block also executing massive corporate staff reductions earlier in February 2026 [1]. For Meta’s remaining employees, the juxtaposition of impending layoffs and the mandatory training of autonomous AI replacements creates a uniquely tense corporate environment [2].