ai
Meta Pauses Employee-Data AI Training After Internal Exposure, Says No Misuse Found
Meta halted its Model Capability Initiative after internal data including keystrokes and conversations became accessible company-wide. The company classified the incident as SEV 2 and said it is investigating while maintaining no evidence of improper access by staff.
Audio version
Tap play to generate a narrated version.
Developing·Limited corroboration so far. This page will refresh as more sources emerge.
Meta paused its internal Model Capability Initiative after employee keystroke and mouse movement data became accessible across the company. The program, announced in April, requires most staff to participate and feeds their activity into AI model training. A screenshot obtained by Business Insider showed the exposed material included private conversations, performance records, and transcriptions.
Meta rated the incident SEV 2 on its internal scale of 0 to 5, where 0 denotes the highest severity. A Meta spokesperson said the company designed the program with privacy safeguards and has no indication that any data was improperly accessed by employees. The spokesperson added that Meta is pausing the training while it investigates the exposure.
Employees expressed frustration in internal messages. One wrote on Monday that the incident left them incensed. Another stated they saw no evidence of malicious access yet found the failure to lock down the data as promised super frustrating.
The pause follows an earlier security incident in which a flaw in Meta's AI chatbot allowed multiple Instagram accounts to be hijacked last month.
