Meta paused its Model Capability Initiative after the internal AI training program inadvertently exposed sensitive employee data. This program collected granular activity data, including employee keystrokes and mouse movements. The leaked information contained private conversations, performance reviews, and transcriptions.
Meta classified the security breach as a high-severity SEV 2 event. A company spokesperson confirmed the program remains suspended pending a full investigation. The spokesperson stated there is currently no evidence that other employees improperly accessed the exposed data.
The initiative previously faced significant internal backlash regarding its mandatory data collection. Staff members raised formal privacy concerns before the leak occurred.