Meta pauses AI training program tracking employee keystrokes after internal leak
6 hours ago
- #AI training
- #privacy concerns
- #data breach
- Meta has paused its internal AI training program, the Model Capability Initiative (MCI), after sensitive data became accessible across the company.
- The leaked data included employees' private conversations, performance data, and transcriptions, classified as a SEV 2 incident on a 0-5 scale.
- Meta confirmed the incident and is investigating, stating there's no indication of improper access but pausing the program as a precaution.
- The MCI program, launched in April, uses staff keystrokes and mouse movements to train AI models, sparking backlash from employees uncomfortable with data recording.
- Employees expressed frustration over the leak, criticizing that data wasn't locked down as promised, with one calling it 'super frustrating'.
- This incident follows other recent security issues at Meta, including an AI chatbot flaw hijacking Instagram accounts and a rogue AI agent incident in March.