Meta taps employee activity data to boost AI development efforts

Meta Platforms is stepping up its investment in artificial intelligence, turning to employee activity data to improve how its systems learn and perform everyday tasks.

The move highlights a growing shift across Silicon Valley, where companies are building AI tools that can handle routine office work with limited supervision. 

The company is introducing software that records how staff use their computers, capturing details such as mouse movements, clicks and keystrokes.

Executives say this type of real-world data is essential to making AI systems more practical and reliable. 

Meta, the parent company of Facebook and Instagram, sees the effort as part of a broader plan to weave AI into its daily operations.

The goal is not only to improve its technology but also to rethink how work gets done inside the company. 

At the same time, the initiative is drawing attention to long-standing concerns about privacy and monitoring in the workplace.

As companies gather more detailed data to train AI systems, questions about how that data is used are becoming harder to ignore.

Building smarter AI systems 

Meta said in internal communications that the new software, known as the Model Capability Initiative, will run on work devices in the United States.

It tracks how employees interact with work-related apps and websites and occasionally captures screenshots to give its AI models more context. 

The company said the aim is to help its systems better understand tasks that humans perform naturally, such as navigating menus or using keyboard shortcuts.

Employees were told they can contribute to improving AI simply by going about their daily work.

Reshaping work with AI 

The project is part of a wider program, now called the Agent Transformation Accelerator, focused on building AI agents that can handle large portions of office work.

Chief Technology Officer Andrew Bosworth said the long-term plan is for these systems to carry out tasks, with employees guiding and reviewing the results. 

A Meta spokesperson said the data collected will be used only to train AI models and not to assess employee performance.

The company added that measures are in place to limit access to sensitive information.

Rising concerns over monitoring 

Meta’s approach reflects a broader trend in the tech industry, where companies are adopting AI tools to write code, process data and streamline operations.

The shift is also influencing hiring, with Meta preparing to cut about 10% of its workforce while investing more in AI-focused roles. 

Still, the use of tracking software has raised concerns among legal experts, who say constant monitoring could tilt the balance of power further toward employers.

While U.S. laws place few limits on workplace surveillance, stricter rules in Europe, including the General Data Protection Regulation, could restrict similar efforts there.