Meta spied on its employees with a keylogger to train AI

Meta recorded every keystroke, conversation, and mouse movement of its employees without giving them the option to opt out. This was the scandal that rocked the artificial intelligence industry in 2026.

In April 2026, Meta launched the internal Model Capability Initiative (MCI): software installed on corporate laptops that captured keystrokes, audio, and private conversations to train its AI models. More than 1,500 employees signed a petition against it. The program continued unchanged. What did stop it was an internal leak that exposed 45,000 tables of sensitive data to any employee at the company—a configuration error that left exposed exactly what they had promised to protect.

This case reveals how far Big Tech companies are willing to go in the race to dominate generative AI, and how the very engineers building these mass-surveillance systems ended up becoming their first victims. If you’re interested in understanding the dark side of AI, subscribe and like this episode so you don’t miss the next ones.