Meta plans to monitor its workers’ every keystroke and mouse click on company computers, using the data to train advanced artificial intelligence systems, the company has confirmed.
In an internal announcement on Tuesday, Meta revealed details of a new tool called the Model Capability Initiative, or MCI, first reported by Reuters. The software will run quietly on staff laptops and internal applications, gathering authentic examples of human computer use to help build AI agents that handle everyday digital tasks.
A Meta spokesperson told the BBC, “If we’re building agents to help people complete everyday tasks using computers, our models need real examples of how people actually use them.” He added that the data is not used for any other purpose and that the tool has “safeguards in place to protect sensitive content.”
The initiative has sparked unease at a time when employees fear significant job losses. Meta has already cut around 2,000 positions this year in targeted rounds, and sources expect larger reductions ahead. One anonymous staff member described the monitoring as very dystopian, telling the BBC, “This company has become obsessed with AI.” A recent former employee called it just the latest way they’re shoving AI down everyone’s throat.
A hiring freeze begun last month has since expanded dramatically. Meta’s careers website, which listed nearly 800 jobs in March, now shows only seven.
While Meta could previously review employee activity for security purposes, this systematic logging specifically for AI training represents a new frontier. Comparable efforts are underway at rivals like Amazon and Microsoft to harness internal data for AI gains.
CEO Mark Zuckerberg has doubled down on the strategy, pledging heavier AI spending this year and declaring 2026 the year that AI dramatically changes the way we work. “We’re starting to see projects that used to take big teams now be accomplished by a single, very talented person,” he said in January.
Meta forecasts $140 billion in AI investment for 2026, nearly double last year’s outlay, following its $14 billion majority stake in Scale AI last year. The first product from its Meta Superintelligence Labs, the Muse Spark model, launched last month, with MCI data set to drive future releases.
Analysts at Gartner predict such tactics could shorten AI development by 30 to 40 per cent, but recent Blind surveys indicate 62 per cent of Meta staff feel AI-related burnout.