Meta is currently holding off on continuing its internal AI training program, as a security risk exposed employees' activity data to other stakeholders within the organization.
The project started this year with the purpose of gathering employee data such as mouse movements, clicks, key strokes, and screen activities. The employee tracking initiative was supposed to help Meta train its AI in understanding people and how they use computers in day-to-day activities.
The problem arose after an internal review of the project revealed that sensitive data was accessible throughout the organization. It is said that employee conversations, AI prompts, and other data related to work were available to unauthorized employees.
This raised concerns about data privacy and pushed Meta to terminate the project until a full investigation reveals how such a leak occurred. The firm noted that there have been no cases of data misuse, but a review of the security measures used in this project is underway.
The incident highlights a major issue that technology firms are increasingly grappling with: user privacy and surveillance.