Meta Employee AI Training

Meta just announced it will track employee screens and keystrokes to train its own AI tools.

This isn't a surveillance story. It's a data strategy story.

Meta is spending $115–135 billion on AI infrastructure in 2026. The hardware is the easy part. The harder problem is proprietary behavioral data — the kind that teaches AI how real professionals actually work, not just how they say they work.

Your keystrokes are a training set. Your workflow patterns are a dataset. Your daily decisions, in aggregate, are a model.

The question every company should be asking right now isn't "should we do this?" It's "what is the value of our employees' behavioral data, who owns it, and what are we obligated to disclose?"

Employment law hasn't caught up. AI governance frameworks haven't caught up. And most employees have no idea this kind of data collection is even possible, let alone happening.

Meta may be first to make headlines on this. They won't be the last.

What obligations should companies have to employees whose behavior becomes AI training data?