‘Very dystopian’: Meta to track employee keystrokes to train AI systems

Meta staff express unease about new data grab

Meta is introducing new internal software to monitor how employees use their computers, as part of an effort to improve its AI systems, according to company documents seen by Reuters.

The tool, known as the Model Capability Initiative (MCI), will record mouse movements, clicks and keystrokes across work-related applications and websites. It will also periodically capture screenshots of employees' screens.

The company says the data will help its AI better understand how humans interact with computers, particularly in tasks such as navigating menus or using keyboard shortcuts.

In a memo shared with staff, Meta said the initiative would allow employees to contribute to improving its AI models "simply by doing their daily work".

Meta's chief technology officer, Andrew Bosworth, told employees in a separate communication that the company is accelerating efforts to develop AI systems capable of performing workplace tasks autonomously.

"The vision we are building towards is one where our agents primarily do the work and our role is to direct, review and help them improve," he wrote, adding that human roles would increasingly involve directing and refining AI outputs.

The initiative forms part of a broader internal programme now called the Agent Transformation Accelerator (ATA).

A Meta spokesperson said the data collected through MCI would be used solely for training AI models and not for employee performance evaluation.

Safeguards, the company said, are in place to protect sensitive information, though details were not specified.

"If we're building agents to help people complete everyday tasks using computers, our models need real examples of how people actually use them — things like mouse movements, clicking buttons, and navigating dropdown menus," the spokesperson said.

Staff concerns over surveillance

Some employees, however, have expressed unease.

One worker described the move as "very dystopian", particularly amid expectations of further job cuts.

Another former employee said the tool reflected a growing obsession with AI inside the company.

Meta has already announced plans to cut around 10% of its global workforce starting in May, with the possibility of further reductions later this year.

Internally, the company has been encouraging staff to rely more heavily on AI tools, even if doing so slows productivity in the short term.

It has also begun restructuring roles under a broader "AI builder" designation.

Tech companies are investing heavily in AI systems capable of automating complex tasks.

Meta's chief executive, Mark Zuckerberg, has pledged to significantly increase spending on AI, with plans to invest roughly $140 billion in 2026, nearly double the previous year's figure.

The company has also expanded its AI teams and recently launched a new model, Muse Spark, through its Meta Superintelligence Labs division.

Experts say the initiative raises important questions about workplace surveillance.

Ifeoma Ajunwa, a ⁠law professor at Yale University, noted that while monitoring tools have historically been used to detect misconduct, tracking detailed user behaviour marks a significant escalation.

"On the US side, federally, there is no limit on worker surveillance," she said, although some states require employers to inform staff when monitoring is taking place.

By contrast, stricter regulations in Europe could make similar practices unlawful.

Valerio De Stefano, a law professor at York University in Toronto, said such monitoring would likely breach data protection rules in many jurisdictions, including provisions under the GDPR.

Stefano added that increased surveillance can shift workplace power dynamics further in favour of employers.