The Model Capability, which records mouse movements and keystrokes to train Meta's AI, has been put on hold after it was revealed that sensitive employee data was accessible across the company.
Meta has suspended a program that tracks its own employees while they work. On Monday, June 22, 2026, the company announced that it would stop operating the internal tool, which tracks employees’ mouse movements and digital activity to train its AI models. At the same time, it is investigating how sensitive employee data ended up being readable by anyone at the company./
Model Capability Initiative operation and concerns
The tool, known as the Model Capability Initiative (MCI), was launched in April 2026. It records the mouse movements, clicks, and keystrokes of employees in the US, with occasional screenshots, and feeds that data into Meta’s models as training data. The goal is to teach AI systems how employees navigate tasks. However, this resulted in a significant amount of personal information being exposed.
The suspension follows revelations in documents reviewed by Reuters that sensitive employee data was accidentally made accessible to all Meta staff. The exposed material reportedly included private conversations, performance data and voice recordings.
See also: Meta suspends cooperation with Mercor after breach
The irony is remarkable: a program designed to monitor employees failed to keep the data safe from the very employees it was monitoring. The exposure was not due to an external breach but to a permissions issue, an internal misconfiguration that turned a set of monitoring data into an open file, for which Meta is responsible.
“We have carefully designed this program with privacy safeguards in place. And while we have no indication at this time that any data was improperly accessed by Meta employees, we are suspending it while we investigate the situation,” spokeswoman Tracy Clayton said. The company did not specify how long the suspension will last.
See also: Italy: Lawsuits against Meta and TikTok after teenager's death
Reactions from Meta employees
MCI was controversial within Meta from the start, and not just because of security concerns. Employees objected to being monitored by software built to learn from them, a tension that intensified when it was introduced before layoffs. To mitigate the backlash, Meta later offered a pause button that allowed employees to disable the monitoring for 30 minutes at a time.
See also: Meta adds AI assistant and desktop version to Edits app
Privacy violation
The legal implications were significant. Recording employees' keystrokes and screenshots conflicts with Europe's data protection regulations, and the program has been flagged for potential violations of the GDPR, which sets high standards for the processing of personal data and considers consent in the workplace problematic due to the imbalance of power between employer and employee.
A leak that made sensitive files widely accessible is precisely the failure that such regulations aim to prevent.
For now, the metrics are no longer being tracked, at least not by MCI. Meta has said it will investigate but has not indicated how long that will take or whether the program will return to its current form, a redesigned version, or not at all. Those responses, as well as the data that prompted the suspension, remain unpublished.
