Internal security notices reveal that potentially sensitive data from Meta employees’ laptops was left accessible to anyone within the company. The data, collected as part of a controversial initiative aimed at training artificial intelligence models, reportedly includes keystrokes and mouse clicks.

The incident has sparked internal debate and questions over privacy reviews. Andrew Bosworth, Meta’s chief technology officer, acknowledged misconfigurations in access control lists (ACLs) and promised to investigate further. Meanwhile, the company is pausing the data collection program indefinitely.

Earlier this year, more than 1,600 Meta employees signed an internal petition protesting against the laptop surveillance effort, citing potential security risks and a lack of sufficient safeguards. The tracking was defended by executives as necessary for training AI systems to mimic human use of software. However, widespread employee protest led Meta to offer more exemptions.

The data breach comes at a time when morale is already low due to mass layoffs and reorganization efforts. In March, Meta introduced the Applied AI team with 6,500 new roles focused on improving AI models, some of which employees found menial and soul-crushing.