A year ago, China's brain–computer interface (脑机接口) field mostly shared enthusiasm, not definitions. Different labs called the same signal by different names; one team's "high channel count" was another's "moderate." That looseness is now being deliberately closed — not by slowing the technology, but by writing the rules underneath it.
The 2026 push is about turning a sprawling research frontier into a regulated industry.
Key takeaways
- A 2026 draft guide (征求意见稿) from the Ministry of Industry and Information Technology plans 40-plus brain–computer interface (脑机接口) standards by 2028 and 80-plus by 2030.
- The guide organizes applications into six categories: medical, industrial, transportation, health, education, and fusion.
- In September 2026 the NMPA released a third BCI medical-device standard that writes AI into the quality rules for brain-electrodata datasets.
- China's BCI market is forecast to reach about ¥12 billion (≈ US$1.7B / HK$13B) by 2030, up from roughly ¥3 billion (≈ US$420M / HK$3.3B) in 2025.
- Legal scholars argue current laws suffice for now, with dedicated legislation deferred until the field matures.
From terminology to a system
The first step was vocabulary. In September 2025 the National Medical Products Administration released standards including Terms for Medical Devices Using Brain–Computer Interface Technology, which took effect on 1 January 2026 and gave the field a shared dictionary. Useful, but narrow — it defined words, not performance or safety.
The 2026 move is broader. The MIIT Equipment Industry Department organized the National Brain–Computer Interface Industry Standard System Construction Guide (2026 Edition) (国家脑机接口产业标准体系建设指南(2026版)), currently a draft for comment (征求意见稿). Its stated targets: formulate or revise more than 40 standards by 2028, and build a basically formed system with 80-plus standards by 2030. The draft groups applications into six buckets — medical, industrial, transportation, health, education, and fusion — so that safety, data, and ethics rules can be tailored per use.
Critically, this is a draft open for comment, not a finalized binding rule. Presenting it as "released policy" would be wrong; the honest status is "published for consultation."
The standard that puts AI in the loop
The more concrete 2026 development is a finished rule. In September 2026 the NMPA issued Brain–Computer Interface Medical Devices — Quality Requirements and Evaluation Methods for Brain-Electrodata Datasets Used in Artificial Intelligence Algorithms (脑机接口医疗器械 用于人工智能算法的脑电数据集质量要求与评价方法). It is described as China's third BCI medical-device standard, and its significance is that it bakes AI directly into the quality bar: how brain-data is collected, labeled, and stored now has a uniform yardstick.
That matters because modern brain–computer interface (脑机接口) decoders are machine-learning models, and a model is only as good as its training data. Without agreed dataset quality, every lab trains on slightly different, uncomparable data — which slows the very "neural large models" the industry talks about. A shared data standard is the unglamorous plumbing that lets cross-hospital research and model training actually happen.
Why standards now
The timing is not accidental. China named brain–computer interface (脑机接口) in its 15th Five-Year planning and the 2026 Government Work Report, and a 2025 seven-ministry opinion set 2027 and 2030 industrial targets. Standards are the connective tissue: they turn scattered research grants into a procurement- and reimbursement-ready industry. As one industry analysis put it, the guide's job is to "give products an identity and draw red lines for the industry" — clarifying which products are medical (strictly regulated) and which are consumer (barred from medical claims).
The market incentive is real. The China Academy of Information and Communications Technology (中国信息通信研究院) projects the domestic brain–computer interface (脑机接口) market at about ¥3 billion (≈ US$420M / HK$3.3B) in 2025 and roughly ¥12 billion (≈ US$1.7B / HK$13B) by 2030. Those figures are forecasts, not results, but they explain why both capital and regulators are moving now rather than later.
The governance question
A recurring question is whether brain–computer interface (脑机接口) needs its own dedicated law. Legal scholars cited in state media argue the current stack — the Product Quality Law, the Data Security Law, and the Measures for the Administration of Science and Technology Ethics Review (Trial) — is sufficient for now. Their suggested path is graduated: use norms, departmental rules, and national and industry standards first, apply differentiated supervision for invasive versus non-invasive and medical versus consumer products, and only pursue dedicated legislation once practical experience accumulates.
That means informed consent, privacy protection, and clarity on liability and damages should be built into standards and ethics guidelines now, rather than waiting for a single sweeping law.
Honest limitations
The MIIT guide is a draft for comment (征求意见稿), not enacted regulation; its 2028 and 2030 standard-count targets are the issuer's own planning figures, not audited commitments. The six application categories and the NMPA September 2026 dataset standard are confirmed via the MIIT published guide and state legal-media reporting; the ¥3B/¥12B market figures are CAICT forecasts reported by state media, not realized market data. The legal scholars' views are commentary, not official policy. "Differentiated supervision" and any future dedicated legislation remain proposals, not enacted rules. No specific company financials are stated here, so the only currency figures are market forecasts, shown with USD/HKD equivalents. This is a policy explainer, not investment or legal advice.
What readers can do now
- For medtech firms, track the NMPA standard family as it expands beyond terminology toward performance and dataset quality — that is where compliance cost and moats form.
- For researchers, engage with the open-comment draft; standards bodies are still shaping definitions you will be measured against.
- For patients and clinicians, watch which products clear medical-device classification — that, not marketing, signals when brain–computer interface (脑机接口) becomes covered, accountable care.