Now Meta will track what employees do on their computers to train its AI agents

Editor's take

Meta has begun monitoring employee computer activity within work applications and web browsers to gather data for training its AI agents. This initiative, dubbed Model Capability Initiative (MCI), is being deployed on US-based employees' machines.

This move is significant as it represents a direct integration of internal workforce data into the development of AI agents, potentially influencing their capabilities in practical, real-world scenarios. It raises questions about employee privacy and the ethical considerations of leveraging proprietary work activity for AI model enhancement, particularly as companies like Microsoft and Google also explore similar internal data utilization strategies.

Future developments to monitor include the specific types of data collected, the clarity of Meta's privacy disclosures to employees, and the ultimate performance improvements observed in their AI agents. The success and perception of MCI will also hinge on whether it leads to tangible benefits for employees or primarily serves as a surveillance mechanism.