{"rewrite":{"id":"r_ce51ec2899a116363b285e32","clusterId":"c_1a2662a6ad8ae3dc88e4be21","slug":"meta-s-brain2qwerty-v2-reads-sentences-without-surgery","model":"deepseek-v4-flash","headline":"Meta's Brain2Qwerty v2 Reads Sentences Without Surgery","summary":"Meta and the Basque Center on Cognition, Brain and Language have developed Brain2Qwerty v2, an AI model that decodes text from non-invasive magnetoencephalography recordings. The new version reads at the word and sentence level, achieving up to 78% word accuracy. Meta also released training code and data.","whyItMatters":"Brain2Qwerty v2 demonstrates that non-invasive brain-computer interfaces can approach the accuracy of surgical implants, moving toward practical communication aids without medical risk.","webCardHtml":"\u003cp\u003eMeta and the Basque Center on Cognition, Brain and Language (BCBL) have released Brain2Qwerty v2, an AI model that decodes typed sentences from non-invasive magnetoencephalography (MEG) recordings. Unlike earlier systems that required surgically implanted electrodes, Brain2Qwerty v2 reads brain activity from outside the skull. The model was trained on roughly 22,000 sentences from nine volunteers who each typed for ten hours while their brain magnetic fields were measured. Brain2Qwerty v2 reads at the word and sentence level, a step up from the character-by-character decoding of the February 2025 v1. Meta reports a character success rate of 69% and word accuracy up to 78%. For the best-performing subject, more than half of all sentences were decoded with an error of one word or less. Accuracy improved with more training data. Meta has also published the training code for both v1 and v2, along with the v1 training data, on GitHub and Hugging Face. The v1 paper, announced in February 2025, passed peer review and was published in Nature Neuroscience on June 29, 2026.\u003c/p\u003e","blueskyPost":"Brain2Qwerty v2 hits 78% word accuracy from non-invasive brain recordings. The leap is not the percentage but that it works at sentence level without surgery, which opens a direct path to communication aids for locked-in patients.","twitterPost":"Brain2Qwerty v2 decodes sentences from non-invasive brain scans at 78% word accuracy. The surgical barrier is gone.","threadsPost":"Brain2Qwerty v2 reads sentences from non-invasive MEG recordings with 78% word accuracy. The previous version worked at character level; this one handles full sentences. Meta also released the training code and data, which means the bottleneck is no longer the algorithm but the hardware cost of MEG machines.","newsletterBlurb":"Meta and BCBL released Brain2Qwerty v2, an AI model that reads typed sentences from non-invasive MEG recordings. The model achieves up to 78% word accuracy without surgery. Meta also published training code and data.","attributionJson":"[{\"source\":\"GIGAZINE\",\"url\":\"https://gigazine.net/news/20260630-brain2qwerty-v2-meta/\",\"title\":\"Meta Develops 'Brain2Qwerty v2,' an AI Model That Reads Text from Brain Activity Without Surgery\"}]","lintFlagsJson":null,"lintHits":0,"costUsd":0,"inputTokens":4253,"outputTokens":610,"status":"published","repairAttempts":0,"nextRepairAt":null,"factsAttemptedAt":1782787254,"createdAt":"2026-06-30T02:36:40.000Z","publishedAt":"2026-06-30T02:39:50.000Z","updatedAt":"2026-06-30T02:39:50.000Z"},"cluster":{"id":"c_1a2662a6ad8ae3dc88e4be21","canonicalTitle":"手術不要で脳活動から文章を読み取るAIモデル「Brain2Qwerty v2」がMetaによって開発される","representativeArticleId":"a_a44749ab193f654d04b130d2","sourceCount":1,"writtenSourceCount":1,"writeAttempts":0,"isSolo":true,"entitiesJson":"{\"anime_titles\":[],\"manga_titles\":[],\"work_titles\":[],\"studios\":[],\"people\":[],\"type\":\"news\",\"domain\":\"other\",\"is_roundup\":false}","contentType":"news","status":"published","firstSeenAt":"2026-06-30T01:59:00.000Z","lastSeenAt":"2026-06-30T01:59:00.000Z","updatedAt":"2026-06-30T02:39:50.000Z"},"attribution":[{"source":"GIGAZINE","url":"https://gigazine.net/news/20260630-brain2qwerty-v2-meta/","title":"手術不要で脳活動から文章を読み取るAIモデル「Brain2Qwerty v2」がMetaによって開発される"}],"entities":{"anime_titles":[],"manga_titles":[],"work_titles":[],"studios":[],"people":[],"type":"news","domain":"other","is_roundup":false},"keyFacts":["Meta and the Basque Center on Cognition, Brain and Language released Brain2Qwerty v2, an AI model that decodes typed sentences from non-invasive magnetoencephalography recordings.","The model was trained on roughly 22,000 sentences from nine volunteers who each typed for ten hours.","Brain2Qwerty v2 achieves a character success rate of 69% and word accuracy up to 78%.","For the best-performing subject, more than half of all sentences were decoded with an error of one word or less.","Meta published the training code for both v1 and v2, along with the v1 training data, on GitHub and Hugging Face."]}