Meta Now Monitors Employee Computers to Train AI Agents

Meta has initiated a new internal program, the ‘Model Capability Initiative’ (MCI), designed to collect employee activity data for the express purpose of training its artificial intelligence models.

This initiative involves installing a specialized tool on the US-based computers of Meta employees. As reported by Reuters, the MCI tool operates within work-related applications and websites, meticulously logging mouse movements, clicks, keystrokes, and capturing occasional screenshots of employee interactions.

The core objective of collecting this data is to enhance Meta’s AI agents, enabling them to better understand and replicate human-computer interaction. This training aims to empower the AI to automate various work tasks, mirroring the functions performed by Meta’s employees. Importantly, Reuters also noted that the data gathered through MCI will not be utilized for employee performance assessments.

Explaining the rationale, Meta spokesperson Tracy Clayton stated to The Verge, “If we’re building agents to help people complete everyday tasks using computers, our models need real examples of how people actually use them — things like mouse movements, clicking buttons, and navigating dropdown menus.” She further elaborated, “To help, we’re launching an internal tool that will capture these kinds of inputs on certain applications to help us train our models. There are safeguards in place to protect sensitive content, and the data is not used for any other purpose.”

The broader strategy for increased internal data collection, including the MCI, was outlined in an internal memo shared with employees on Monday by Meta CTO Andrew Bosworth. This data will contribute to Meta’s Agent Transformation Accelerator (ATA). Bosworth articulated a future vision where, “The vision we are building towards is one where our agents primarily do the work and our role is to direct, review and help them improve.”