Eight years ago at F8, Facebook announced its ambition to develop a device that could type using brain waves, aimed at helping ALS patients like the late Dr. Stephen Hawking communicate without a physical keyboard. Now, Meta has finally achieved a breakthrough, developing a “mind-reading machine” with up to 81% accuracy.

Created by Meta’s Fundamental AI Research (FAIR) team, the system reads brain activity without requiring implants, unlike Neuralink. Instead, it uses Magnetoencephalography (MEG) machines, which capture brain waves at 1,000 frames per second, and processes the data through an AI model called Brain2Qwerty. In a study conducted in Spain, 35 volunteers had their brain activity scanned while typing, allowing AI to predict their words based on thought patterns.

However, the technology is far from being ready for public use. MEG machines are large, expensive, and require magnetically shielded rooms, making real-world application impractical for now. Additionally, accuracy varies, with only some volunteers achieving 81% accuracy, while the overall average stands at 68%.

Meta plans to continue refining the system, working toward the vision of a thought-powered typing device without implants, as promised eight years ago.