Robotics labs have spent years recording humans doing physical tasks to teach their systems when and how to grab, walk, or stack boxes. Meta just brought that playbook to software and computer use, except the demo subjects are its own employees. The program is called Model Capability Initiative. It captures mouse movements, clicks, and keystrokes across hundreds of apps and websites including Google, LinkedIn, GitHub, and Slack. No opt-out. The data goes straight into training AI agents that can perform knowledge work autonomously.
My first reaction is that this is genuinely creepy. Recording every keystroke and click of your workforce, with no opt-out, to train the systems that might eventually replace them, is about as dystopian as it gets. The backlash inside Meta has apparently already started.
Then again, when you strip out the AI training framing, companies have always run analytics on their employees. Productivity dashboards, email monitoring, time tracking software. What Meta is doing is arguably just the next version of that, except the output is a model instead of a spreadsheet. I really hope I’m wrong about where this leads, but I can see a world where “training data collection” becomes the new justification for a level of workplace surveillance that would have felt unacceptable five years ago.