Meta is converting the daily computer operations of its employees into AI training data, a move that pushes tech giants' bets on artificial intelligence to a new dimension—not only replacing human labor with AI but also &#34;extracting&#34; human effort to feed it.
On April 22, according to Reuters, an internal memo revealed that Meta has begun deploying tracking software on work computers of U.S. employees, collecting real-time mouse movement trajectories, click behaviors, and keyboard operations, while periodically capturing screen content; all data will be used to train its AI models. The tool, named the &#34;Model Capability Initiative&#34; (MCI), operates across work-related applications and websites.
This deployment is part of Meta's broader &#34;AI Transformation&#34; strategy. Andrew Bosworth, Meta's Chief Technology Officer, stated in another internal memo released on Monday that the company is intensifying internal data collection efforts, aiming to build AI agents capable of &#34;primarily undertaking work tasks,&#34; while human employee roles will shift to &#34;commanding, reviewing, and assisting improvements.&#34; The strategic project has been renamed the &#34;Agent Transformation Accelerator&#34; (ATA).
<h2>Feeding AI with Employees' Daily Operations: The Logic Behind the MCI Tool</h2>
The memo indicated that the MCI tool was launched by an AI research scientist from Meta SuperIntelligence Labs, a model-building team under Meta, on an internal channel on Tuesday.
The memo explained that the tool aims to address AI models' shortcomings in simulating human-computer interactions—for example, selecting options from dropdown menus or using keyboard shortcuts. Currently, AI models still struggle to replicate these nuances of natural human behavior.
&#34;This is precisely where all Meta employees can help advance the models through their daily work,&#34; the memo read.
Meta spokesperson Andy Stone confirmed that data collected by MCI would serve as one of the inputs for AI training, stating that the data would not be used for employee performance evaluations or any other purpose. Protective measures have been implemented to filter out &#34;sensitive content,&#34; though specific types of excluded data were not detailed. Stone further explained:
<blockquote>
&#34;If we are to build agents that help people complete daily computer tasks, the models need authentic samples of human operations—such as mouse movements, button clicks, and navigation within dropdown menus.&#34;
</blockquote>
Reports indicate that the deployment of MCI is a microcosm of Meta's large-scale AI transformation strategy. Internally, the company has already begun requiring employees to use AI agents to complete tasks such as programming, even if it temporarily reduces efficiency. Meanwhile, Meta is breaking down functional boundaries between certain roles, promoting a general position known as &#34;AI Builder.&#34;
Last month, Meta established a new Applied AI (AAI) engineering team focused on enhancing the programming capabilities of AI models and building AI agents capable of handling product construction, testing, and release tasks. Earlier this month, Meta began transferring &#34;high-performing&#34; software engineers into the AAI team.
In his memo, Andrew Bosworth painted a vision: &#34;The future we are building is one where agents primarily undertake work, while our role is to command, review, and help them improve.&#34; He added that the goal is for agents to &#34;automatically identify areas where they feel human intervention is needed, so they can perform better next time.&#34;
<h2>White-Collar Monitoring Controversy: Legal Boundaries and Power Imbalance</h2>
This initiative has sparked significant controversy regarding legal boundaries and labor rights.
Yale Law School Professor Ifeoma Ajunwa pointed out that computer logging and screenshot technologies have historically been used by companies to investigate employee misconduct or non-work activities. However, recording keyboard operations in this instance pushes data collection to a deeper level, subjecting white-collar employees to real-time monitoring previously reserved for delivery drivers and gig economy workers.
&#34;At the federal level in the United States, there are no restrictions on employee monitoring,&#34; Ajunwa stated, noting that state laws merely require employers to provide broad notice when monitoring employees.
In Europe, the legal landscape differs significantly. Valerio De Stefano, a law professor at York University in Toronto specializing in technology and comparative labor law, stated that such monitoring would likely be deemed illegal in Europe.
In Italy, using electronic monitoring to track employee productivity is explicitly prohibited; in Germany, courts have ruled that employers may deploy keyloggers only in special circumstances, such as suspected serious criminal activity. Furthermore, De Stefano believes this practice may also violate the EU's General Data Protection Regulation (GDPR).
De Stefano also noted that the rising awareness among employers regarding monitoring has shifted the power dynamics in the workplace on a macro level, tilting the balance further in favor of employers.

METD

METU

According to reports, Meta is collecting mouse movements, keyboard operations, and screenshots from U.S. employees to train AI, aiming to build AI agents capable of performing core tasks while transforming human roles into "supervision and review." This deep monitoring has sparked labor rights and legal controversies; while unrestricted in the U.S., it may violate laws in Europe, exacerbating power imbalances in the workplace

- Meta has begun deploying tracking software on employees' work computers to collect data on their mouse movements, clicks, and keyboard actions for AI training.  
- This initiative, part of Meta's broader AI transformation strategy, aims to create AI systems capable of undertaking work tasks, while adjusting human roles to supervising and refining these systems.  
- The move has stirred legal and labor rights concerns, particularly regarding the extent of surveillance and its implications for employee monitoring rights in the U.S. compared to stricter regulations in Europe.

"Refined Employees"? Meta Collects Employee Mouse Movements and Keyboard Operations to Train AI