Meta suspends employee monitoring tool after data leak reveals keystrokes and private information

Meta confirmed that a monitoring program deployed in April recorded mouse movements, clicks, keystrokes and even private conversations, performance reviews and health data from its US‑based staff. The data were intended for training the company’s internal AI models, but the tracking software was classified as a SEV‑2 security incident when internal documents showed that sensitive employee information could be viewed by other staff.

Meta said there was no evidence of unauthorized external access, but the company is investigating the lapse and has begun to pause the tool, with a gradual shutdown planned. Employees have protested the practice, describing it as a violation of privacy. The incident follows other recent Meta security flaws, including a vulnerability that allowed takeover of Instagram accounts and an earlier AI‑agent mishap that exposed internal data.

Company spokesperson Tracy Clayton emphasized stricter access controls for the raw data and noted that only a limited number of engineers have direct access. The tracking program remains active for some users as the phased suspension proceeds.