Meta to track workers' clicks and keystrokes to train AI
Meta to track workers' clicks and keystrokes to train AI
Meta plans to log employee computer usage—mouse clicks and keystrokes—to train its AI models, with safeguards and no other purposes.
Meta will start tracking the way employees work, logging their computer activity to help train its artificial intelligence models. The internal tool, which will run on Meta's computers and apps, will record actions such as mouse movements, clicks, and keystrokes to generate training data for AI systems. The goal is to give models real examples of how people actually use computers to build agents that can assist with everyday tasks. Data collected will be used solely for training and improving models, with safeguards in place to protect sensitive content. The company says the information will not be used for other purposes and that privacy protections are in place.
Employees have reacted with mixed feelings. Some worry the move signals a broader push toward automation and AI adoption at the expense of workers, calling it dystopian or a pathway to more job cuts. Meta has already laid off about 2,000 employees this year in smaller rounds, and a recent hiring freeze appears to be intensifying. Listings on Meta's public job site have dropped dramatically—from around 800 in March to just seven now—raising questions about the company’s staffing plans.
The tracking tool at the heart of the effort is described as the Model Capability Initiative. By collecting real-user interactions, Meta aims to improve the autonomy and reliability of its AI agents. While the company asserts that the data is strictly for AI training and protected by safeguards, the move underscores the broader trend of embedding AI into everyday work processes—and the accompanying concerns over privacy and workforce impact. Meta has not publicly commented on the job-listing reductions, but the shift highlights the balancing act between AI advancement and employee trust.
