Meta shut down a program that monitored employee keystrokes and mouse movements on June 22, 2026, after a security breach leaked employee data, Reuters reported. The pause came after months of workers pushing back against the tool.

The program started in April 2026. Meta wanted to collect data on how employees typed and moved their mice on work computers, so the company could use that data to train artificial intelligence systems, Reuters reported. The idea: AI systems learn better when they have lots of real human behavior data. Employees at Meta were a handy, available source.

But employees were not happy. Over 1,600 workers signed a petition against the program, and some held a protest in May 2026, Reuters reported. The timing was bad for Meta's case: the protest happened just before the company announced it would lay off 10% of its workforce. People who might lose their jobs did not want the company watching how they worked. They saw the tracking as spying, not research.

In early June, Meta tried to fix the problem by letting employees opt out. That did not work, Reuters reported. Then the data leak happened, and Meta decided to stop the entire program.

Europe's privacy rules blocked the path forward

Employee anger was not Meta's only problem. The European Union has strong privacy laws called GDPR that limit what companies can do with employee data. Tracking what workers type falls into a category of risky surveillance under these rules, Reuters reported. European regulators take employee privacy very seriously. If Meta wanted to keep the program, it would have had to prove the tracking was absolutely necessary—a high bar to clear.

This matters for other tech companies too. Many are looking for ways to gather data on how people behave to improve their AI systems. But if they operate in Europe, they all hit the same EU privacy wall.

What happens next is still up in the air

Meta said it is pausing the program while it investigates the security breach. It did not say the program is gone for good. The data leak forced Meta to stop for now, but bigger questions remain: Can the program be fixed and restarted? Will regulators allow it?

The 1,600 workers who signed the petition made their position clear. But Meta has found ways around employee complaints before. The real blocker could be European regulators, who might demand the program be changed so much it would barely be worth doing.

Here is the broader issue: companies want employee behavior data to train AI because it is genuinely useful. But getting that data from workers inside a company is harder than getting it from the internet. Employees have legal protections, can organize, and can refuse to participate—unlike the data Meta scoops up from the public online. Meta's experience shows that even the world's largest tech companies hit limits.