Meta has introduced a new AI system that translates a person’s brain activity into written text without requiring brain implants or surgery.
Called Brain2Qwerty v2, the technology marks one of the biggest advances yet in non-invasive brain-computer interface research. It could eventually help people who have lost the ability to speak communicate more easily.
“The system uses external brain scans instead of implanted devices. This brings accuracy closer to methods that previously required invasive procedures.”
-said company officials
The research is part of Meta’s wider Digital Brain Project and has been released alongside open-source code to support further scientific work.
Also Read: Meta Rolls Out AI-Powered Search on Facebook, Reshaping How Users Discover Content
How does Brain2Qwerty v2 work?
Brain2Qwerty v2 depends on magnetoencephalography (MEG), a non-invasive brain imaging technique that records the tiny magnetic signals produced by the brain while a person types. These signals are then processed by an end-to-end AI model, which reconstructs the intended words and sentences.
Unlike earlier systems that decoded characters one at a time, the latest model predicts complete words and sentences. Meta trained the AI using around 22,000 sentences collected from nine volunteers, each of whom spent roughly 10 hours inside an MEG scanner while typing.
Also Read: Meta Teams Up With Reliance for First AI-Enabled Data Centre in India, Bets Big on Jamnagar
Accuracy improves over previous methods
As quoted by Meta, Brain2Qwerty v2 achieved an average of 61 per cent word accuracy, while the best-performing participant reached 78 per cent.
Previously, the non-invasive approach achieved 8 per cent word accuracy, highlighting the significant improvement.
The company also reported that, in its strongest results, more than half of an individual’s sentences were decoded with one word error or fewer. Meta said the technology now approaches the performance levels previously reserved for systems that require brain surgery.
Also Read: Meta Bets Big on Humanoid Robots for Homes After a Startup Acquisition
Research tool, not a consumer product
Meta stressed that the technology remains a research project rather than a commercial product. The current system depends on large, highly specialised MEG scanners that are expensive and confined to laboratory environments, making everyday use impractical.
The company believes the long-term goal is to develop communication tools for people affected by neurological conditions, paralysis or brain injuries. Meta has also made the research code publicly available to encourage wider collaboration across the neuroscience community.
Brain2Qwerty v2 reshapes AI development, moving beyond chatbots & image generators into healthcare and neuroscience.
While commercial use is years away, the research signals growing investment in brain-computer interfaces and could shape the next wave of AI-powered medical innovation.
Samarjit Kaur is a journalist and communications professional covering technology & emerging digital trends. With a focus on clarity and context, she reports on developments shaping industries and governance. When not reporting, she chooses to plug-in and relax on her playlists and plan her next bucket-list trips!
