Meta is launching the "Model Capability Initiative" to record US employees' computer interACTions for AI training, rAIsing questions about workplace privacy and job security.
The US tech conglomerate Meta is taking a controversial step to integrate Artificial Intelligence into the workplace by harvesting the behavioral data of its own employees. Following a year that has already seen workforce reductions, the company has notified its US-based staff of a new tracking tool designed to log their daily computer activities.
The Model Capability Initiative (MCI)
According to internal mEMOs reviewed by Reuters, Meta is deploying a tool called the Model Capability Initiative (MCI) on the computers of its US-based full-time employees and contractors. This software is designed to capture granular data regarding how workers interact with their machines.
The tracking includes:
Mouse movements and CLIcks: Recording the trajectory and location of cursor interactions.
Keystrokes: Logging keyboard inputs.
Screen Content: Taking occasional screenshots to provide visual context for the actions.
The initiative covers a pre-APProved list of work-related applications and websites, including Gmail, GChat, internal tools like Metamate, and coding platforms like VSCode. Meta has clarified that the tool applies only to work computers, not personal mobile devices.
training the Next Generation of AI Agents
Meta states that the primary purpose of the MCI is to train its next-generation AI models, specifically to help "Agents" underStand how humans perform everyday tasks on a computer.
While AI Models excel at research and technical Skills like coding, Meta argues they lack an understanding of basic computer interactions, such as navigating drop-down menus or using keyboard shortcuts. The data collected will be used to train Muse Spark, a frontier-scale system launched in April by Meta’s Superintelligence Labs. By mimicking these human interactions, Meta hopes to build agents that can autonomously handle complex, multi-step professional tasks.
"If we are building agents to help people complete everyday tasks using computers, our models need real examples of how people actually use them," a Meta spokesperson stated.
Employee Backlash and Privacy Fears
The rollout of the MCI has been met with significant resistance and anxiety among Meta's workforce. Employees have described the move as "dystopian," expressing discomfort at the prospect of constant surveillance.
Key concerns include:
Job Security: With Meta planning further layoffs—potentially affecting up to 10% of its global workforce—employees fear they are effectively training the AI agents that will replace them.
Privacy: Despite assurances, workers worry about the security of sensitive data, including passwords and personal Information, being captured by the tracking software.
In response to the backlash, Meta spokesperson Andy Stone emphasized that strict safeguards are in place. He assured employees that the data would not be used for performance evaluations or any purpose other than model training. However, the company has made it clear that opting out of the program is not an option for US staff.
A Shift in Workplace Dynamics
This initiative marks a significant shift in corporate surveillance, reminiscent of the monitoring often seen in the gig Economy but now applied to high-level white-collar roles. It reflects CEO Mark Zuckerberg's vision for 2026 as the year AI Dramatically transforms work, moving towards a structure where AI agents handle the bulk of execution while humans shift to supervisory roles.
As Meta invests roughly $140 billion in AI infrastructure this year, the line between employee and training data source is becoming increasingly blurred, signaling a new era where workplace privacy may be the price of technological advancement.
Comments & Questions (0)
No comments yet
Be the first to comment!