Meta has stopped using a program that monitored what employees typed on their computers and how they moved their mice, after security problems exposed this information. The decision came in late June 2026, according to Reuters and Wired, following a formal security alert filed by an employee.

Why Meta Started Tracking

Meta launched the program in April 2026, installing software on computers used by U.S. employees. The company said it wanted to collect this behavioral data to help train artificial intelligence systems.

Employees pushed back immediately. By mid-May, workers organized protests at Meta's offices, handing out leaflets and asking colleagues to sign a petition. For a company not known for visible employee organizing, this was significant.

In early June, Meta made a small change: employees could pause the tracking for 30 minutes at a time, and those who objected could request not to participate. But this did not satisfy the concern — and then a more serious problem surfaced.

What Went Wrong

An employee filed a formal security alert after discovering that the data the program collected had been exposed or accessed improperly. The details of exactly what happened are not yet public, but the significance is clear: keystroke data can reveal a lot. It might contain passwords, private messages, or patterns showing which employees did what.

Think of it this way: there is a big difference between recording someone's general work habits versus recording every single letter they typed. One is broad statistics. The other is a detailed record of their actions.

Most companies have strong rules about protecting customer data — your password, your messages. But many have not created the same safeguards for data they collect about their own employees. Meta's pause gives the company a chance to build those protections, though it is not yet clear whether it will do so or simply restart the program after the immediate problem is solved.

Why This Happened Now

The tracking program was not the only major change at Meta in recent months. On May 20, 2026, Meta announced plans to reassign 7,000 employees to AI work and eliminate a management position. Employees already worried about their jobs were now being asked to accept being monitored at their desks — a difficult combination.

When you introduce keystroke monitoring across thousands of computers without building proper security first, problems are likely. The security alert that shut down the program suggests that basic protections were not in place when the tracking started.

Other technology companies face similar pressure. They want to gather data that helps them build better AI systems, and employee data is one source. But the difference between a program that works and a program that creates trouble comes down to how carefully you set it up, who has access to the information, and what rules you put in place to protect it.

Meta's program is paused for now. What happens next — whether the company fixes it, makes it smaller, or stops it entirely — is still an open question.