Latest / Health and life sciences

Model releaseHealthScienceFrontierUS SpainConfirmed

Meta releases Brain2Qwerty v2, decoding typed sentences from non-invasive brain recordings

See how this story connects

Meta AI released Brain2Qwerty v2, a non-invasive system that decodes brain activity into text using magnetoencephalography (MEG) recordings and end-to-end deep learning. It was trained on approximately 22,000 sentences from nine volunteers who typed while recorded, and reached 61% word accuracy, with 78% for the best participant. Meta released the full training code for v1 and v2, and the Basque Center on Cognition, Brain, and Language released the v1 dataset.

Why it matters

Meta says the research could help the millions of people whose brain lesions prevent them from communicating; its non-invasive system reached 61% word accuracy against 8% for other non-invasive methods, approaching levels previously exclusive to surgical techniques.

When
29 Jun 2026
ID
WTA-N-2026-0404
Who
Meta AI, Basque Center on Cognition, Brain, and Language

Summary by wheresthe.ai, from Meta AI. Checked against the source on 24 Oct 2026.

Source copy: download the page text as saved on 24 Oct 2026 (PDF). Also held by the Internet Archive.

Next stop on this Storyline: What happened nextMeta launches Muse Image model with agentic capabilities

Cite this story

WTA-N-2026-0404: “Meta releases Brain2Qwerty v2, decoding typed sentences from non-invasive brain recordings.” wheresthe.ai, 29 Jun 2026. https://wheresthe.ai/d/us-meta-releases-brain2qwerty-v2-decoding-typed-sentences-fr-2026-06/ (source: Meta AI).

Spotted an error? Email [email protected].