Meta is installing new software on its employees' computers in the U.S. to monitor mouse movements, clicks, and keystrokes. The data collected will be used to train AI models, according to Reuters, citing internal documents.

This initiative is part of a broader program aimed at developing AI agents capable of performing work tasks autonomously.

The tool is called the Model Capability Initiative (MCI) and operates within work-related applications and websites. It is also configured to periodically take screenshots.

The goal of the initiative is to enhance models in areas where they perform poorly, such as making selections from dropdown menus and using keyboard shortcuts.

“This is where all Meta employees can help our models improve simply by doing their everyday work,” the documents state.

Meta's Chief Technology Officer Andrew Bosworth noted in a separate memo that the company will intensify its internal data collection as part of the AI for Work initiative.

“We aim for agents to do the work while you guide and review them, helping them improve,” the executive added.

Meta spokesperson Andy Stone assured that the data collected through MCI will not be used for performance evaluations or any purposes other than training models. Security measures are also in place to protect confidential content.

Automation and Unemployment

AI tools allow for the automation of certain tasks and reduction of personnel costs, which some companies are taking advantage of.

Meta plans to lay off 10% of its workforce starting May 20, with additional significant cuts being considered by the end of the year.

Amazon has terminated contracts with 30,000 workers over the past few months. In February, Block cut nearly half of its staff.

Meta encourages the use of AI agents for programming and other tasks, even if it slows down work in the short term.

Last month, the company formed a team of engineers focused on Applied AI. Their task is to enhance the programming capabilities of neural networks and use them to create agents. These agents will eventually be employed in the development, testing, and launch of future products and infrastructure for the corporation.

In early April, Meta began transferring its top programmers to the Applied AI team.

Legality of the Initiative

Yale Law School Professor Ifeoma Ajunwa noted that tracking computer activities and taking screenshots have historically been used by companies to identify employee misconduct.

The shift to tracking keystrokes elevates data collection to a new level. Employees are subjected to a degree of real-time surveillance previously experienced only by couriers and gig economy workers.

“There are no federal restrictions on employee monitoring in the U.S.,” Ajunwa stated.

York University law professor Valerio De Stefano emphasized that European legislation would likely prohibit such monitoring.

It is worth noting that in April, Meta introduced the AI model Muse Spark, developed by a new team from Meta Superintelligence Labs.