Source Credit : Portfolio Prints
Meta is rolling out new tracking software on U.S.-based employees’ computers to capture mouse movements, clicks, and keystrokes—data it plans to use to train its artificial intelligence systems. The initiative, detailed in internal memos seen by Reuters, is part of a broader push to develop AI agents capable of autonomously performing workplace tasks.
The tool, known as Model Capability Initiative (MCI), will operate across work-related applications and websites, periodically capturing screen snapshots as well. According to one memo shared internally within Meta’s AI research teams, the goal is to generate real-world interaction data to help models better mimic how humans use computers.
The company says its AI still struggles with everyday digital behaviors—such as navigating dropdown menus or using keyboard shortcuts—and believes passive data collection from employees’ routine workflows can close that gap. As one internal note put it, employees can help improve AI systems simply by doing their daily work.
Meta Platforms—the parent company of Facebook and Instagram—has been aggressively embedding AI into its operations, positioning automation as central to improving efficiency and reshaping its workforce.
In a separate memo, CTO Andrew Bosworth said the initiative, now branded as the Agent Transformation Accelerator (ATA), reflects a long-term vision where AI agents handle most operational tasks, while humans shift toward oversight roles—directing, reviewing, and refining machine output.
Bosworth emphasized that building such systems requires large-scale, high-quality data on real workplace interactions. While he did not detail specific training pipelines, he noted the company would be “rigorous” in collecting and evaluating interaction data across workflows.
Meta spokesperson Andy Stone confirmed that data collected through MCI would feed into model training but said it would not be used for employee performance evaluations. He added that safeguards are in place to exclude sensitive information, though specifics were not disclosed.
The move reflects a wider trend across the U.S. tech sector, where companies are accelerating automation efforts. AI tools have demonstrated growing capability in handling complex tasks—from software development to data organization—fueling both investor enthusiasm and concerns about workforce displacement.
Meta itself is planning to cut roughly 10% of its global workforce starting May 20, with further reductions under consideration. Meanwhile, Amazon has reduced its corporate headcount by tens of thousands in recent months, and Block has also undertaken significant layoffs.
Internally, Meta is encouraging employees to adopt AI agents for tasks like coding—even at the cost of short-term productivity losses. The company is also restructuring roles, replacing specialized job titles with a more generalized “AI builder” designation aligned with its automation-first strategy.
As part of this shift, Meta recently launched a new Applied AI (AAI) engineering team focused on improving the coding abilities of its models and deploying AI agents capable of building, testing, and shipping products with minimal human intervention. The company has already begun reallocating top engineering talent into this unit.
However, the expanded use of monitoring technology is raising legal and ethical concerns. Historically, workplace tracking tools have been used to detect misconduct or ensure productivity, but logging detailed keystroke data marks a deeper level of surveillance.
Ifeoma Ajunwa noted that such practices extend real-time monitoring—previously more common among gig and delivery workers—into white-collar environments. Under U.S. federal law, there are few restrictions on employer surveillance, though some states require basic disclosure.
In contrast, similar practices would likely face legal challenges in Europe. Valerio De Stefano pointed out that strict regulations, including the General Data Protection Regulation (GDPR), would likely prohibit this level of monitoring. In countries like Italy, tracking employee productivity through electronic surveillance is explicitly banned, while German courts allow keystroke logging only under exceptional circumstances.
Beyond legal considerations, experts warn that pervasive monitoring could shift workplace power dynamics further toward employers, raising broader questions about privacy, autonomy, and the future of human work in an AI-driven economy.
