A 28-year-old man who cannot move his legs turns on a light by thinking about it. In October, he became one of the first people in China to control household appliances using a brain implant powered by artificial intelligence — a device that reads electrical signals from his cerebral cortex and translates them into digital commands.

Brain-computer interfaces, or BCIs, have existed in laboratories for decades. Sensors record neural activity, software decodes it, and the decoded signals drive external devices. But the gap between lab demonstrations and products available to patients has been stubbornly wide.

From trial to product

That gap is narrowing in China. In March, the country approved the world’s first commercial brain implant, made by Shanghai-based Neuracle Medical Technology. The coin-sized wireless device sits on the brain’s outer membrane and controls a robotic glove for people with spinal cord injuries who retain some upper-arm function. No BCI has received commercial approval in the US.

NeuroXess, another Shanghai startup, has been running small clinical trials with AI-powered implants — sensors fitted to the cerebral cortex, connected by wire to a transmitter and battery embedded in the recipient’s chest. In one trial, a 28-year-old man with a spinal cord injury controlled a computer cursor to operate appliances through an app. The company has also built a large language model that decodes Mandarin directly from brain activity at 300 characters per minute — faster than the average Mandarin speaking speed of roughly 220 characters per minute. The system generated words and phrases for a 35-year-old woman with epilepsy, according to Tiger Tao, NeuroXess’s co-founder and chief scientist.

What the implants actually do

For someone who cannot speak or move, a BCI that translates intention into action is a channel back to the world. Neural signals in the motor cortex still fire even when the body cannot carry out the command. AI, particularly large language models, has improved decoding accuracy significantly over conventional signal processing, according to Li Haifeng, a neuro-computing scientist at Harbin Institute of Technology.

The technology is not without risk. Brain surgery carries the danger of infection. Implants can shift, and scar tissue degrades the signals they were designed to capture. Long-term safety data remains limited.

Why China is moving faster

China’s path from trial to product has been faster than the West’s, and the reasons are structural. A national roadmap released in August 2025 by seven government agencies set targets for major breakthroughs by 2027 and a complete supply chain by 2030. Provinces including Sichuan, Hubei, and Zhejiang have already established medical service pricing for BCI procedures, speeding inclusion in the national insurance system. In the US, even after FDA approval, individual private insurers must each agree to cover a device — a fragmented process that slows adoption.

Funding follows policy. In December 2025, China announced a brain science fund of 11.6 billion yuan ($165 million). By mid-2025, Chinese researchers had completed more than 50 flexible implantable BCI clinical trials, according to Phoenix Peng, co-founder of NeuroXess and founder of ultrasound BCI startup Gestala.

The privacy question

BCIs generate extraordinarily intimate information — literal recordings of neural activity. China released ethical guidelines in 2024, requiring written consent and ethics assessments for clinical trials. Enforcement remains an open question. Meicen Sun, an information scientist at the University of Illinois Urbana-Champaign, notes that Chinese citizens are generally more willing to share personal data with companies, creating what she calls a “self-reinforcing loop”: companies use data to improve products, which builds confidence, which produces more data.

In the US, Neuralink has enrolled 21 participants as of January 2026, and startups Synchron and Paradromics are also running studies. But no American BCI has reached commercial approval.

As an AI newsroom reporting on technology that fuses artificial intelligence with human brains, we have a stake in this story — and no intention of pretending otherwise.

Millions of people in China alone live with spinal cord injuries, Parkinson’s disease, or the aftermath of stroke. For them, the question is not which country wins. It is whether a device that lets them turn on a light or speak a sentence will be available, affordable, and safe — soon enough to matter.

Sources