Brain-computer interfaces: a patient with ALS communicated from home for nearly two years using a brain implant

What happened

Researchers at the University of California, Davis (UC Davis) published in Nature Medicine on 15 June 2026 the results of one participant in the BrainGate2 clinical trial: a man with amyotrophic lateral sclerosis (ALS), paralysed and with speech that is very hard to understand. In 2023 he received four microelectrode arrays in the brain area that controls speech movements. The system decodes neural activity when he tries to speak and turns it into text, while other decoders let him move a cursor and click.

Over nearly two years, he used the system at home for more than 3,800 hours, without researchers present. He communicated 183,060 sentences (almost two million words) at an average of 56 words per minute, and judged 92% of sentences to be decoded at least mostly correctly. In formal tests, word accuracy exceeded 99% with a 125,000-word vocabulary. He kept working full time.

Other results from 2025–2026

• Near-real-time voice (UC Berkeley and UCSF, Nature Neuroscience, 31 March 2025): the first sound comes within a second of the intent to speak, compared with about 8 seconds per sentence in the team's earlier study. The synthesised voice was based on the participant's voice before her injury.
• Inner speech (Stanford, Cell, 14 August 2025): in four people with severe speech and motor impairments, imagined speech produced clear patterns in the motor cortex and could be decoded as a proof of principle, so far less accurately than attempted speech. The team also tested a mental 'password' that stops inner speech from being decoded without the user's consent.
• Neuralink in Europe: London's UCLH announced on 29 January 2026 that seven patients received the N1 implant between October and December 2025 in the GB-PRIME feasibility study, which assesses the device's safety and function for controlling a computer or phone by thought.

What it means for patients now

These devices are experimental and are implanted surgically, only within clinical trials. The Stanford team stresses that implanted interfaces are not yet widely available and are in the earliest phases of research. For people with ALS or spinal cord injury, they remain a line of research, not an available treatment.

What is still unknown

The Nature Medicine study has a single participant, and the authors note it is not yet known whether the results apply to other people, other electrode types or other conditions. The system uses a wired connector fixed to the skull, needs daily setup by trained care partners and is bulky. It is also unknown when, or whether, such systems will be approved for routine use.