Research Explainer · XiaoHu Explains
No surgery needed to read brain signals — Meta's non-invasive system hits nearly 8x the accuracy of prior methods
Nine volunteers wore an MEG device while typing; Meta reports 61% word accuracy and released v1/v2 training code plus the v1 dataset
⚑Stance note: this article is based on Meta AI's official blog post. The 61%, 78%, etc. accuracy figures are Meta's own self-reported results across 9 participants; the 8% comparison baseline is cited from other research (Nature Neuroscience).
1What It Is
MEG signals become text during a controlled typing task
In June 2026, Meta AI reported that Brain2Qwerty v2 could decode sentences in real time from MEG (magnetoencephalography) recordings during a controlled typing task, and released training code for v1 and v2.
Primary source and reference
- Meta AI Blog Original publisher