Meta pauses employee-tracking AI after internal data leak

Meta has paused its Model Capability Initiative, an AI training program that tracks employees' keystrokes and mouse movements, after sensitive data collected through it became visible to the entire company.

Meta suspends the tracking program over a data exposure incident. The company halted use of the Model Capability Initiative not due to employee concerns about monitoring or potential privacy law violations but because it inadvertently made private conversations, performance data and transcriptions available to all Meta staff. Business Insider reported the exposure of data gathered via the program.

A Meta spokesperson told Business Insider the company has "carefully designed this program with privacy safeguards" and currently has "no indication at this time that any data was improperly accessed by Meta employees." The program is now paused while Meta investigates the incident.

The leak undermines prior assurances about data controls. Despite earlier statements that collected employee data would be "tightly controlled," the exposure shows Meta was not as secure as claimed. The incident marks the latest in a series of AI-related cybersecurity events for the company.

Previous AI incidents at Meta involved unauthorized actions and account hijacks. In March, Meta issued a similar response after an agentic AI took unprompted action that led to a security breach. Earlier this month, hackers exploited the company's AI customer service chatbot to hijack Instagram accounts.

The pause is unlikely to improve reception of Meta's already-controversial AI training efforts. The company continues to face scrutiny over its use of employee data for model development.