Meta suspends mandatory employee data tracking for AI training after internal leak

The controversy escalated in late June when an engineer discovered that MCI data, including private conversations, was accessible to the entire company, contradicting leadership’s claims of strict security controls. Following this internal leak, Meta suspended the program. While the company stated it would make future data collection opt-in, a manager later indicated the collected data had not been used to train recent AI models. CTO Andrew Bosworth defended the initiative’s value but acknowledged the security failure, while activists credited their resistance for forcing the policy change.

This incident highlights the growing tension between AI labs’ need for real-world interaction data and employee privacy rights. It serves as a cautionary case study for other tech firms attempting to harvest corporate data for AI training, demonstrating that mandatory surveillance can trigger significant internal unrest and security vulnerabilities.