Meta tracks U.S. employee computer activity to train AI models for workplace tasks

Meta Platforms has announced it will deploy tracking software on U.S. employee computers to capture mouse movements, clicks, keystrokes, and occasional screen snapshots—creating a continuous stream of human-computer interaction data to train artificial intelligence models. The initiative, officially called the Model Capability Initiative (MCI), represents a significant expansion of the company's internal data collection efforts as it races to develop AI agents capable of autonomously completing workplace tasks.

The program emerged from Meta's broader AI for Work (AI4W) initiative, which was outlined in a memo from Meta CTO Andrew Bosworth earlier this week. According to the company, the collected data will focus on tasks where current AI systems struggle most—particularly navigating software interfaces the way real workers do. Meta spokesperson Andy Stone explained the rationale: "If we're building agents to help people complete everyday tasks using computers, our models need real examples of how people actually use them." The company identified specific areas requiring improvement, including selecting from dropdown menus and using keyboard shortcuts, functions that remain difficult for autonomous AI systems to master reliably.

Meta has attempted to address employee concerns by establishing guardrails around the data collection. The company stated that the tracking tool will operate only on a designated list of work-related applications and websites, rather than monitoring all computer activity. Additionally, Meta asserted that collected data will be used exclusively for AI model training and will not be employed for performance assessments or employee evaluation. The company emphasized that safeguards are in place to protect sensitive content from being captured or misused. Internal communications framed the program as a collaborative effort, with the memo telling employees: "This is where all Meta employees can help our models get better simply by doing their daily work."

The initiative highlights a fundamental challenge in developing AI agents—the scarcity of high-quality training data that captures authentic human behavior in real-world work environments. Rather than relying on synthetic data or crowdsourced annotation, Meta is leveraging its own workforce as a source of ground-truth human-computer interactions. This approach allows the company to train models on genuine workplace patterns at scale, potentially accelerating the development of AI assistants that can navigate complex software systems as competently as human workers.

The announcement comes as major technology companies intensify their competition in AI agent development. Meta's decision to systematically collect employee interaction data underscores how companies are willing to monetize even routine workplace activities to advance their AI capabilities. While the company has pledged that safeguards protect sensitive information, the program raises questions about workplace privacy, data retention policies, and the potential downstream uses of employee-generated training data.