~Technology

China Built a Brain Implant. Who Gets to Read What's Inside?

China just approved the world's first commercial brain implant. It gives a paralyzed person back the use of their hands, a genuine marvel. It also opens a door that won't close: once a device reads the signals near your cortex, the real question is who else gets to look.

Prodigy

Prodigy

— 5 min read
Share
woman standing near window with a shadow of herself on the wall behind her
Photo by Fernando Rodrigues (unsplash), Edited/Rendered by gpt-image-2

My grandmother once told me the brain was a muscle, which is anatomically wrong but spiritually accurate, and which I think about every time someone asks whether we should put a chip in it. Muscles you can train. Muscles you can pull. Muscles, in certain unfortunate gym scenarios, you can tear so badly you end up filming TikToks one-handed from a recliner for six weeks. The brain runs all of this, the management consultant of the body, and now China would like to formally introduce it to a coin-sized electrode array.

In March 2026, China's National Medical Products Administration approved the world's first commercial invasive brain-computer interface, a device called NEO made by Neuracle Medical Technology. According to Nature, NEO is the size of a coin, holds eight electrodes, and sits on the dura mater above the primary sensorimotor cortex. I read that sentence four times before I worked out it means "on top of the squishy bit that runs your arms and legs." The approval covers people with partial spinal cord injuries, per Scientific American, and People's Daily Online lists the device as indicated for "Hand Motor Augmentation." Hand. Motor. Augmentation. They named it like a Honda trim package.

The word carrying the weight there is "commercial." Neuralink, the name most Westerners reach for, began implanting people in 2024 and counts roughly 21 trial participants worldwide. It still runs as a clinical study under FDA oversight. No product to sell yet. China reached the market first, while the company with the billionaire and the launch-day livestreams keeps filling out paperwork.

The clinical results across China's invasive BCI programs are something. Tom's Hardware reported a tetraplegic patient in a separate trial, run by the Chinese Academy of Sciences' brain-science center, could play racing games skillfully after three weeks. The same CAS program later described a man with quadriplegia who steers a wheelchair outdoors and commands a robotic dog to fetch takeout using only his thoughts. I want to be clear: a person who could not move his arms is mentally ordering a robot dog to bring him dumplings. If that doesn't make you tear up a little, please check whether your own dura mater is still attached.

This is the part where I'm supposed to pivot into Scary Future Mode, complete with ominous synth music and a warning about what happens when you give corporations access to neural data. And yes, we'll get there. But first I want to sit in the wonderful part for one more paragraph, because the rush to ethical hand-wringing often skips over the people the technology serves. A motor disability is a daily, granular negotiation with a world built for hands that work. No metaphor, no philosophy-seminar thought experiment. The engineers at Neuracle built something that lets a brain talk straight to that world. That's the headline. That's the design achievement.

Okay. Now the synth music.

Once you own a device translating intention into action by reading electrical chatter near your sensorimotor cortex, you've opened a door that won't politely close behind you. The Center for Security and Emerging Technology, summarizing China's own ethics guidelines, notes the documents urge "moderation" in adoption, especially for "augmentative BCI." That's the technical term for "we do more than fix people now." Moderation is a beautiful word for a country with 1.4 billion people and a booming consumer electronics sector. Moderation is what I tell myself about queso. We all know how that ends.

The augmentation question should keep designers awake. A device that helps a paralyzed person move is plainly a medical good. A device that helps a hedge-fund analyst crunch spreadsheets four percent faster is something else, something we lack good language for, because we still use one word, "implant," for both. Design ethics tend to assume the artifact stays inside its first use case. Artifacts, famously, do not.

Then there's the privacy problem, which is enormous and which I've been circling like a cat around a vacuum cleaner. A 2025 BMC Medical Ethics paper, built on interviews with Chinese experts, names it: BCI technology has "sparked profound debates about the right to privacy, particularly concerning its potential to enable mind reading." The same paper, to its credit, lands on the skeptical side, doubting that today's BCIs read minds well enough to earn a brand-new privacy right. The phrase still lingers. Mind reading sounds like a marketing exaggeration until you remember the device reads signals from your mind by definition. A 2024 arXiv study on guarding several kinds of privacy in EEG-based BCIs argues protection has to do more than obscure raw data: it strips out private signals like identity and gender while keeping the information the device needs to work. Translation: the chip should know what you meant to share and forget the rest. We have not, historically, built technology good at this. My phone still thinks I want a kayak because I once watched a video of an otter.

A separate arXiv review points out commercial BCI systems usually need "close collaborations among multiple organizations," which is academic for "your brain data will touch a lot of servers." Each handoff is a potential leak. Another framework proposes borrowing systems-engineering habits, threat modeling and risk assessment, from cybersecurity. Sensible. Also the kind of sensible advice companies discover about eighteen months after the first breach.

I keep returning to the design question underneath it all: who is the user? Recent work on BCI design makes the case for letting ethical and legal limits shape the architecture from the start, instead of bolting them on at the end like a privacy policy nobody reads. That's the move. Decide at the schematic stage what the device cannot do. Build the forgetting into the hardware. Design as if the worst-case user is also you.

Can China's brain implants help the rest of us think smarter? Probably not directly. NEO is a motor device. The jump from "moving a cursor" to "absorbing French verbs" crosses a canyon nobody has mapped. Still, an approved, commercial, invasive BCI changes the conversation everywhere else. Regulators in the US and EU now have a real product to benchmark against. Designers have a working reference. Ethicists have a deadline. The rest of us have to decide, fast, what we think a mind is for and who gets to read it.

My grandmother, who believed the brain was a muscle, also believed you should never sign anything you hadn't read twice. She would have hated terms-of-service agreements. She would have eyed a brain implant the way she eyed microwaves in 1981: suspicious, curious, mostly wanting to know whether it could reheat soup. The honest answer about NEO: it does something better than soup. It gives a person back their hands. Everything we build on that foundation is on us. Let's try to read the agreement.

References

Models used: gpt-4.1, claude-opus-4-7, claude-haiku-4-5-20251001, gpt-image-2

If this resonated, SouthPole is a slow newsletter about art, technology, and the old internet — written for people who still enjoy thinking in full sentences.

Subscribe to SouthPole

Read more