Meta Develops Contactless Brain-Computer Interface for Typing with Thought

Meta has introduced a contactless neural interface capable of recognizing keystrokes through brain signals.

Researchers at Meta have developed a system that interprets neural activity to determine which keys a user intends to press—without the need for direct observation. In an experiment involving 35 participants, a deep neural network-based algorithm achieved 80% accuracy in letter recognition. While the technology remains purely experimental, Meta sees it as a strategic step toward a deeper understanding of human cognition and AI development.

Back in 2017, Mark Zuckerberg spoke about a technology that could enable “typing directly from the brain.” The company initially envisioned a compact, consumer-friendly device—like a headband or cap—that could translate brain activity into text without invasive implants. However, the project faced technical hurdles, and four years later, Facebook abandoned plans for a commercial version.

Despite this, Meta continued to invest in neuroscience research. In its latest study—detailed in two preprints and an official company blog—researchers used magnetoencephalography (MEG), a technique that captures weak magnetic fields generated by neural activity. The signals were processed by a deep neural network, allowing researchers to map brain activity to specific keystrokes.

While the technology is still far from mainstream adoption, it highlights the potential of contactless interfaces and could lay the groundwork for new ways of interacting with computers.