Tech giant Meta Platforms is rolling out a controversial new initiative to train its next-generation artificial intelligence systems — by tracking how its own employees use their computers, according to a report.

The move, first reported by Reuters, involves installing specialised tracking software on employee devices in the United States that records mouse movements, clicks and keyboard activity as part of efforts to build more advanced AI models.

Training AI Through Human Behaviour

The data collection forms part of Meta’s broader strategy to develop AI systems capable of performing everyday digital tasks autonomously — from navigating menus to executing commands with human-like precision.

Internal documents indicate the tool will operate on selected work-related apps and websites, and may also capture periodic screen snapshots to better understand user context.

The collected data will be used by Meta’s AI research teams, including its “Superintelligence Labs,” to refine models that still struggle with basic operational tasks such as selecting correct dropdown options or efficiently using keyboard shortcuts.

Company Defends Move

Meta spokesperson Andy Stone said the initiative is aimed solely at improving AI systems and will not be used to evaluate employee performance.

“Understanding how people naturally interact with computers is essential if we want to build AI agents that can operate like humans,” the spokesperson said, adding that safeguards are in place to protect sensitive information.

Privacy Concerns Persist

Despite assurances, the development is likely to raise concerns over workplace privacy and surveillance, particularly as tech firms race to develop more capable AI systems.

Experts note that while such data could significantly improve AI usability, it also blurs the line between innovation and intrusion, especially if monitoring expands beyond limited testing environments.

AI Race Intensifies

The initiative highlights the growing competition among global tech companies to build AI systems that can replicate complex human-computer interactions — a key frontier in the evolution of artificial intelligence.

As companies push the boundaries of what AI can do, the debate over how far they should go in collecting human behavioural data is expected to intensify.