Assistive technology. stories

Jun 29, 2026

Meta AI unveils Brain2Qwerty system that decodes typed text from brain activity without implants

Meta AI researchers have developed Brain2Qwerty, a deep learning model that non-invasively decodes typed sentences from brain activity using external sensors, achieving an 18% character error rate. The breakthrough, published in Nature Neuroscience, represents a significant advance toward non-invasive brain-computer interfaces for communication.

Jun 29, 2026·3 sources

Jun 15, 2026

Brain implant enables ALS patient to communicate for nearly two years, generating 2 million words

Casey Harrell, a 42-year-old man with ALS who lost his ability to speak, has successfully used a brain-computer interface for almost three years to communicate and control his computer, producing approximately 2 million words. The at-home brain implant represents a significant milestone in assistive technology for neurodegenerative disease patients.

Jun 15, 2026·3 sources

Top claims

▪Character accuracy peaks at 22% for MEG and 16% for EEG in linear classification tasks, both significantly above the 14% chance level.
▪Brain2Qwerty was evaluated on MEG recordings from 20 participants comprising 193,000 characters, 30,000 words, and 5,000 sentences.
▪MEG significantly outperforms EEG in Brain2Qwerty decoding performance with a P-value of 1.0 × 10−7.

Subtopics

Neuroscience Deep learning Brain-computer interfaces AI research & benchmarks Medical implants Neurodegenerative disease

Related timelines

Assistive technology. stories

Meta AI unveils Brain2Qwerty system that decodes typed text from brain activity without implants

Brain implant enables ALS patient to communicate for nearly two years, generating 2 million words

Related timelines

US Congress

Venture capital funding

Iran War

Strait of Hormuz blockage

Trump administration