Meta will utilize data from its employees’ mouse movements and keystrokes to enhance its AI models, Reuters reports. This initiative aims to develop more capable and efficient artificial intelligence.
The decision reflects a growing trend among technology companies seeking unconventional sources of training data, which are crucial for improving AI performance. The move raises significant questions regarding data privacy and employee consent.
A Meta spokesperson stated, “If we’re building agents to help people complete everyday tasks using computers, our models need real examples of how people actually use them — things like mouse movements, clicking buttons, and navigating dropdown menus.”
Meta will launch an internal tool designed to collect these inputs from certain applications to aid its AI training efforts. The company confirmed that it has implemented safeguards to protect sensitive content and emphasized that the collected data will not be used for any other purpose.
This trend underscores emerging privacy issues within the AI industry. Recently, reports indicated that older startups are being repurposed for their corporate communications, such as Slack and Jira archives, to create additional training data for AI models.
