A decisive truth, applied directly to the forehead of workplace privacy: Meta has announced it will capture its employees’ mouse movements, clicks, and keystrokes to train AI models. And if you just felt a chill run down your spine, congratulations — your survival instincts still work.

According to Meta, the company needs “real examples of how people actually use” computers to improve its AI agents for task automation. Which is a perfectly reasonable thing to say if you ignore every science fiction movie ever made.

🤚 The Open-Palm Privacy Check

Meta promises “safeguards are in place to protect sensitive content.” And we believe them, because if there’s one company with a flawless track record on user privacy, it’s— well, it’s not Meta. It’s literally never been Meta. But sure, let’s give them the benefit of the doubt while they install digital surveillance on every employee’s computer.

The justification is actually solid from a pure engineering standpoint: if you want to build AI that automates computer tasks, you need data on how humans actually perform those tasks. It’s like wanting to build a robot chef — you’d study real chefs. Except in this case, the chefs didn’t sign up to be studied while they make their lunch.

👐 The Industry-Wide Awkwardness

Meta isn’t alone in this. Atlassian recently caught heat for enabling default data collection for AI training as well. The pattern is emerging: companies need behavioral data, employees have behavioral data, and the space between “asking” and “taking” is getting uncomfortably thin.

🌿 The Gentle Awakening

Here’s the uncomfortable question nobody in Silicon Valley wants to answer: at what point does “training data” become “surveillance”? Is it when they record your keystrokes? Your mouse movements? The 47 seconds you spend staring at the void between emails?

Meta employees reportedly have the option to opt out. But we all know how opt-outs work in practice: technically available, socially impossible, and definitely noted in your performance review under “team player: needs improvement.”

This article was written by a human, keystroke by keystroke. Fortunately, nobody was recording — as far as we know.