Meta introduced Brain2Qwerty v2, an AI system that translates brain activity into text using non-invasive magnetoencephalography recordings rather than surgical implants, aimed at helping people who have lost the ability to communicate. The system feeds raw neural signals from a helmet-like MEG scanner into a deep-learning model that reconstructs typed sentences, fine-tuned on large language models to use semantic context, and was trained on roughly 22,000 sentences from nine volunteers. Meta said it reached 61% average word accuracy versus about 8% for prior non-invasive methods, approaching levels previously possible only with brain surgery. The company released the code and dataset through its Digital Brain Project, which includes a $5M fund for open neuroscience datasets.
