Meta’s employee surveillance program hit by internal data exposure

Sensitive employee data that was collected to train AI models was accessible to anyone from within the company.

In April 2026, Meta announced that it was installing software on computers operated by American employees to collect data to train and build future AI agents.

The tool, also known as the Model Capability Initiative (MCI), was designed to capture mouse movements, clicks, and keystrokes, but also take periodic snapshots of the content on employees’ screens.

The purpose of this was to improve the company’s AI models in areas where they struggle to replicate how humans interact with computers, like choosing from dropdown menus and using keyboard shortcuts.

“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,” Andrew Bosworth, Chief Technology Officer (CTO) and Head of Reality Labs at Meta, said in a statement at the time.

The initiative caused a lot of backlash from workers, who rose against the company’s invasive surveillance policy to train AI models. A Meta spokesperson stated that he was confident in the tool’s privacy measures and assured that the collected data would not be used for performance assessments.

However, a recent internal security notice says that the information collected by Meta’s tool has been leaked. The notice was sent out on Monday. It has been seen by WIRED and has also been confirmed by three current employees familiar with the issue.

According to the note, employee data across 45,000 hive tables has been exposed internally and includes employee activity like prompt transcriptions, private conversations, and performance data.

Tracy Clayton, a Meta spokesperson, has confirmed that the company is investigating the security issue. He added that the data collection program has been stopped.

“We have carefully designed this program with privacy safeguards, and while we have no indication at this time that any data was improperly accessed by Meta employees, we’re pausing it while we investigate,” Clayton told WIRED.

Last month, over 1,600 Meta workers signed an internal petition protesting the surveillance practices, warning that “collecting this data introduces both security and regulatory risks for Meta, including the potential for breaches and unauthorized disclosure.”