Employee backlash forced Meta to scale back its plan to track workers' keystrokes and mouse clicks for AI training, with new controls letting staff pause data collection for up to 30 minutes at a time and request exemptions, according to an internal memo seen by Reuters. The retreat follows weeks of internal uproar over the Model Capability Initiative (MCI), a program launched in April that logged exact keystrokes and screen content on corporate MacBooks to train AI models on how humans perform office tasks. More than 1,500 employees signed a petition opposing the tracking.
Stephane Kasriel, a vice president in Meta's Superintelligence Labs unit, authored the memo announcing the changes. Kasriel said the team introduced "several optimizations" to reduce the software's drain on laptop battery life, a major complaint from workers who reported the tool consumed so much data it caused internet usage surges when working from home.
"While we remain confident in the privacy protections we put in place at launch, which went through several layers of risk review. We have heard your concerns about personal data on work devices, battery life, and wanting more control over when capturing happens," Kasriel wrote. The MCI tool monitors keystrokes and screen content within a pre-approved list of work apps, including Gmail, Google Chat, VSCode, and Meta's internal assistant Metamate. When employees initially raised concerns, Meta CTO Andrew Bosworth told staff there was "no option to opt out on your corporate laptop."
One Meta employee who spoke to the BBC on condition of anonymity described having their actions train AI models as "very dystopian," especially as workers braced for additional job cuts. A former employee called the tracking tool "just the latest way they're shoving AI down everyone's throat."
Meta has laid off about 2,000 employees this year and told staff in April it planned to cut 10% of its workforce, roughly 8,000 people. The company has simultaneously committed over $600 billion to AI-ready data centers by 2028 and poached top AI researchers with compensation packages as high as $300 million.
Meta declined to comment on the record about the policy changes.
