Meta records its employees to train AI and leaks their data

Published on June 26, 2026 | Translated from Spanish

Meta installed software on its employees' computers to record screens, keystrokes, and clicks with the aim of feeding its artificial intelligence models. The program was suspended after a leak exposed internal personnel information. For the public, this new episode confirms that a company with a history of privacy scandals does not even protect its own people's data.

office surveillance software interface on a computer monitor, screen recording indicator active, keyboard and mouse input capture visualization, abstract data streams flowing from the computer into a glowing AI network, Meta logo subtly reflected on the monitor glass, digital padlock icon broken with data fragments scattering, dark corporate cubicle environment, red warning glitch effects on the screen, cinematic cyberpunk aesthetic, photorealistic technical illustration, dramatic low-key lighting with blue and red neon accents, ultra-detailed hardware components, tense atmosphere of exposed privacy breach

How the internal surveillance system worked 🕵️

The tool, designed to capture every employee interaction with their computer, aimed to generate massive datasets for training algorithms. Meta did not clearly communicate the scope of the monitoring or the security protocols. A configuration flaw allowed sensitive data, such as conversations and passwords, to be exposed. The company stopped the program, but the damage to its reputation was already done: not even its own workers are safe.

The boss is watching you, but forgot to lock the door 🔓

It seems Meta believes the best way to innovate is to spy on its own staff. The funny thing is, with so much focus on every click, no one noticed the data door was open. So, while the company plans its next move, employees already know their privacy is the price of having a job. At least the AI will know exactly how many times you looked at the clock today.