Meta's FAIR AI team uses Brain2Qwerty v2 to translate brain activity into typed sentences, with no implants or surgery required. The system reads magnetic signals outside the skull and reconstructs what a person is typing. Clinical use for paralyzed patients is still a long way off, but accuracy keeps improving with every additional recording. AI agents that wrote their own code helped with the optimization.

Meta AI's Brain2Qwerty decodes typed text from non-invasive brain signals with a 32% character error rate using MEG, but real-time operation remains

Meta says its latest Brain2Qwerty system translates brain activity into sentences using non-invasive brain recordings.