META JUST CROSSED THE LINE: Your Boss Is Now Tracking Every Click and Keystroke to Train AI — And You're Next

META JUST CROSSED THE LINE: Your Boss Is Now Tracking Every Click and Keystroke to Train AI — And You're Next

April 22, 2026 — Mark Zuckerberg just made the dystopian workplace a reality. In a move that should send a chill down the spine of every worker on the planet, Meta has launched a sweeping employee surveillance program that tracks every single keystroke, every mouse click, every action its workers take on company computers — and it's feeding all of that data directly into its AI training pipelines.

This isn't a rumor. This isn't speculation. Meta confirmed it to the BBC. The program, called the Model Capability Initiative (MCI), is actively running on Meta's internal systems right now. And while Meta claims it's only for building "agents to help people complete everyday tasks," the implications are staggering — and terrifying.

Because here's what nobody at Meta wants you to think about: if they're doing this to their own employees today, they'll be doing it to YOU tomorrow.

--

On Tuesday, Meta informed its global workforce about a new tool that would be installed on all company computers and internal applications. The tool's purpose, as described by a Meta spokesperson, was straightforward: "If we're building agents to help people complete everyday tasks using computers, our models need real examples of how people actually use them."

In other words: we're going to watch everything you do, record all of it, and feed it to our AI.

Meta added a single, almost laughable assurance: "The data is not used for any other purpose" and that there are "safeguards in place to protect sensitive content."

Let me translate that corporate doublespeak for you:

This is the company asking you to trust them with a complete recording of every digital action you take at work. The same company whose former employees have testified before Congress about internal knowledge of harms. The same company that Cambridge Analytica proved could be weaponized with the data it collects.

And now they want your keystrokes.

--

Let's be clear: workplace monitoring isn't new. Companies have tracked employee computer usage for decades. Keyloggers, screen capture, email monitoring — these have been standard in corporate environments for years, particularly in regulated industries.

But what Meta is doing is fundamentally different. It's not just monitoring for compliance or productivity. It's harvesting human behavior at the granular level to replicate and replace human workers with AI.

Think about what a keystroke log actually contains:

Meta isn't just collecting data about WHAT you do. They're collecting data about HOW you think. And they're using that to build AI systems that can replicate your cognitive processes.

A Meta employee, speaking anonymously to the BBC, didn't mince words: "This company has become obsessed with AI." Another person who recently left Meta called the tracking tool "just the latest way they're shoving AI down everyone's throat."

Employees described the feeling as "very dystopian" — having their smallest computer actions harvested to train the AI that may eventually replace them.

And here's the kicker: they're not wrong to be worried.

--

To understand how serious Meta is about this transition, look at the investment numbers.

Meta plans to spend roughly $140 billion on AI in 2026. That's nearly double its 2025 AI spending. To put that in perspective:

And where is that money going?

The message is unambiguous: Meta is going all-in on AI, and human workers are the raw material being fed into the machine.

Every keystroke Meta's employees type is being used to refine AI agents that will eventually eliminate the need for those same employees. Meta is literally having its workers train their own replacements — and charging them their privacy as the price of continued employment.

--

The MCI program represents something unprecedented in workplace surveillance: the mass harvesting of cognitive behavioral data for AI training purposes.

Traditional workplace monitoring tracks whether you're working. This tracks HOW you think.

Every time an employee writes a function, debugs code, drafts an email, designs a graphic, or analyzes data, they're generating a rich signal about human problem-solving patterns. Meta is capturing all of it.

And while Meta claims the data is "not used for any other purpose," history suggests otherwise:

Meta has a documented pattern of finding ways to monetize and operationalize every scrap of data it collects. The idea that keystroke logs will remain strictly confined to "training agents to help people complete everyday tasks" requires a level of trust that Meta has done absolutely nothing to earn.

Even if you trust Meta's current leadership — and you absolutely should not — what about the next leadership? What about when Meta's behavioral datasets are subpoenaed in litigation? What about when hackers breach Meta's training data repositories? What about when a future acquisition or partnership gives third parties access to insights derived from this surveillance?

The data Meta is collecting cannot be un-collected. The AI models being trained on it cannot be untrained. If this dataset ever leaks, is misused, or falls into the wrong hands, the detailed behavioral profiles of thousands of workers will be permanently exposed.

--

Meta says there are "safeguards in place to protect sensitive content." Let's examine what that actually means in practice.

We don't know what the safeguards are because Meta hasn't disclosed them. But based on industry standards and Meta's own track record, here's what "safeguards" typically look like:

Meta was fined $5 billion by the FTC for privacy violations. It paid $725 million to settle a class-action lawsuit over unauthorized data sharing. It faced investigations on multiple continents for its handling of user data.

And now we're supposed to believe they'll handle complete behavioral surveillance with appropriate care?

--

Perhaps the most insidious aspect of Meta's MCI program is how it normalizes a level of workplace surveillance that would have been unthinkable just a few years ago.

When keyloggers and screen monitoring were introduced in the 1990s, there was widespread outrage. Employee unions protested. Privacy advocates sued. It was understood as a fundamental violation of worker dignity.

Today, those tools are standard in many industries — and nobody bats an eye.

Now Meta is pushing the boundary again. Total behavioral capture for AI training. And because it's framed as "building helpful agents" and because it only affects Meta employees (for now), the outrage is muted.

But this is how surveillance always spreads:

Meta has already proven they can execute this playbook. They did it with social media tracking. They did it with facial recognition. They're doing it now with workplace surveillance.

The question isn't whether this will spread beyond Meta. The question is how long it takes.

--