Train employees with AI! Meta launches an internal tracking tool that records all employee mouse movements and keystrokes

Meta installs internal tracking tools on US employees’ computers to record mouse, keyboard movements, and screenshots, collecting data to train AI models. The company emphasizes that this does not involve performance evaluations, but it still sparks employee monitoring and privacy concerns, with European regulations potentially banning such practices.

Meta launches internal tracking tools to train AI with employee behavior

Meta is installing Model Capability Initiative (MCI) internal tracking tools on the computers of US employees. This software runs on work-related applications and websites, recording employee mouse movements, clicks, keystrokes, and periodically capturing screenshots.

The purpose of collecting data is to train its own AI models, enabling AI to better mimic human computer operations.

A Meta spokesperson told Reuters that, the data collected by MCI will never be used for employee performance evaluations or for any purposes other than model training.

However, while Meta claims to have taken measures to protect sensitive content, it has not specified which types of data will be excluded from collection.

The US does not restrict “white-collar surveillance,” but Italy and Germany have strict regulations

Yale Law Professor Ifeoma Ajunwa states that recording keystrokes takes data collection a step further, exposing white-collar employees to real-time monitoring previously only experienced by delivery workers. However, the US government has no restrictions on labor monitoring; at most, state laws require employers to inform employees about monitoring.

Toronto York University Law Professor Valerio De Stefano points out that European laws are very likely to prohibit such monitoring practices. Italy explicitly states that using electronic surveillance to track employee productivity is illegal; in Germany, courts have ruled that employers can only deploy keystroke logging tools in cases of suspected serious criminal activity.

Image source: Negative Space free stock library, showing a woman white-collar worker working on a Mac computer (illustration)

Meta doubles down on AI, reshaping internal work models

According to BBC, Meta CEO Mark Zuckerberg recently pledged to increase AI project spending, aiming to invest about $140 billion in AI by 2026.

Zuckerberg has predicted that 2026 will be a year of significant change in work due to AI, with the company heavily investing in AI technology. Besides Zuckerberg himself trying to code with Claude, internally, Meta has launched the Token Legend leaderboard, recording employees’ token usage on AI tools as a performance metric.

• Related report: Meta doubles down on AI: Zuckerberg codes with Claude, employees launch token consumption battles to hit KPIs

Meta’s Chief Technology Officer Andrew Bosworth mentioned in an internal memo that the company will strengthen internal data collection to accelerate AI agent transformation projects.

A recently resigned Meta employee revealed that the internal tracking tools are part of the company’s latest efforts to promote AI. As Meta shifts its funding toward AI R&D, employees expect further layoffs in the future. Recently, Meta decided to cut 10% of its global workforce by the end of May, approximately 8,000 employees losing their jobs.