Meta has made a significant leap in brain-computer interface technology. In collaboration with the Basque Center on Cognition, Brain and Language, researchers at Meta’s Fundamental Artificial Intelligence Research (FAIR) lab successfully decoded unspoken sentences with up to 80% accuracy using brain activity data — all without the need for surgery.
This breakthrough was achieved through magnetoencephalography (MEG), which records the magnetic fields produced by brain cells when they send electrical signals. Unlike invasive brain implants or bulky fMRI machines, MEG offers a non-invasive way to capture brain activity in near real-time.
During the study, 35 volunteers typed sentences while wearing an MEG scanner, resembling a futuristic salon hair-dryer. Researchers trained an AI model to match brain activity patterns with corresponding letters, refining predictions with a language model trained on Wikipedia articles. The AI accurately guessed the intended text even when participants made mental “typos,” effectively autocorrecting their thoughts. 
For a more portable approach, researchers also used EEG electrodes placed on the scalp, decoding the exact letters typed about 33% of the time. While less accurate than MEG, EEG is a promising step toward lightweight, wearable devices.
This advancement brings Meta closer to its vision of mind-reading headbands — a project announced in 2016. Although consumer-ready devices are still years away, wearable MEG scanners already exist, hinting at a future where brain-to-text communication could become mainstream.
However, this innovation raises ethical questions. While mind-reading tech could be life-changing for people with paralysis, concerns about brain data privacy and potential misuse by tech giants like Meta loom large. As the technology evolves, policymakers may need to establish strict regulations to protect users’ mental privacy.
For now, Meta’s research marks a major step toward turning thoughts into text — a development that could reshape how we interact with technology and each other.
