Why Meta is Tracking Staff to Train Workplace AI Agents

Meta is introducing an internal data-collection tool to capture how employees work as it intensifies a push into workplace automation. 

Reuters reports that the move follows layoffs of about 2,000 employees earlier in the year amid a broader refocus on Gen AI.

In April 2026, the company notified US-based staff of the Model Capability Initiative (MCI), which will run on work machines and inside internal apps to log activity.

A new era of surveillance

According to Reuters, MCI will capture keystrokes, mouse movements and occasional screenshots to document how employees use their computers.

Meta says the initiative forms part of a broader effort to build AI agents that can anonymously perform routine workplace tasks.

Several employees express concern that enhanced tracking could enable micromanagement at a time of job uncertainty.

Pandemic-era remote work tools normalised monitoring for many firms, but specific logging to train AI models represents a step-change in scope.

"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 told the BBC.

The company says MCI includes safeguards to protect sensitive content and that the data collected will not be used for any other purpose.

MCI data is expected to support Muse Spark, a frontier-scale system launched in April 2026 by Meta Superintelligence Labs to handle complex, multi-step workflows.

By mimicking human-computer interactions, Muse Spark aims to learn professional office tasks that could underpin future enterprise products.

Workforce impact and investment

Staff say basic device activity has long been accessible internally, but targeted logging for AI training is new.

Andy Stone, Meta spokesperson and Vice President, says MCI data will not be used for performance assessments or for any purpose beyond model training.

Reuters reports internal unease over the prospect of additional cuts as hiring remains tight and the company prioritises AI.

Mark Zuckerberg, Co-founder and CEO of Meta, says 2026 will be "the year that AI dramatically changes the way we work," adding that projects once handled by big teams can now be accomplished by "a single, very talented person".

Mark has revealed Meta plans to spend roughly US$140bn on AI in 2026 – almost double that of 2025. 

Consumer playbook and privacy questions

Meta, which owns Instagram and Facebook, is known for analysing user behaviour to optimise targeted advertising and content.

Some employees are now reporting similar unease inside the company as work activity becomes training fuel for Gen AI.

The shift raises questions about consent, transparency and governance for data collected from employees, even when anonymised.

Meta maintains that MCI will be used solely to improve its models and that privacy safeguards are in place.