Meta Internally Tracks Employee Keystrokes & Mouse Clicks to Train AI Models

Meta is deploying a new internal tool that actively records and converts its own employees' keystrokes and mouse movements into training data for its artificial intelligence models. This move represents a significant escalation in the use of internal workplace data for AI development, directly linking employee computer interactions to the company's core AI research and product pipeline.

The tool, as described by the company, is designed to capture the granular details of how employees interact with their computers—every click, scroll, and keystroke. This raw interaction data is then processed and formatted to serve as training material for Meta's AI systems. While the stated purpose is to improve AI models, the implementation creates a pervasive new layer of workplace surveillance, fundamentally changing the nature of data collection within the corporate environment.

The initiative places Meta at the forefront of a contentious ethical and operational frontier. It raises immediate questions about employee privacy, data consent, and the security of potentially sensitive information captured during routine work. The practice also signals a strategic shift where internal human-computer interaction becomes a proprietary resource, potentially granting Meta a unique and controversial dataset to compete in the intensifying global AI race. The long-term implications for workforce trust, corporate policy, and regulatory scrutiny are now under a new and intense spotlight.