08/26/2026
Your WiFi router can recognize you. No camera, no phone, no login. Just the box blinking on the shelf next to the TV.
Researchers at the Karlsruhe Institute of Technology in Germany recently gathered 197 volunteers and had them walk around a room fitted with standard WiFi hardware, the same kind of gear you'd find in an office or an apartment. Then they asked their system to figure out which of the 197 people was present, using only the radio signals in the air. It got the answer right close to 100 percent of the time.
To understand how, you have to stop thinking of WiFi as a cable without the cable. Your router isn't sending a tidy beam to your laptop. It's filling the entire room with radio waves that bounce off walls, furniture, the ceiling, the dog, and you. Every wave that hits your body comes back a little different, arriving a fraction later, a bit weaker, and bent around the outline of whatever it touched.
Your body is a specific shape. Height, shoulder width, the way you hold your neck, the slight lean you've had since school. All of that changes how the waves scatter, so a taller person bends the signal one way while a heavier person absorbs more of it. Stand ten people in the same spot and the router sees ten different reflections. The researchers call this a fingerprint, and you leave one in the signal every time you walk through a room.
The part that turns this from a lab curiosity into a real problem is a feature called beamforming. Since the WiFi 5 standard arrived, routers have been aiming their signal at your devices instead of blasting it evenly in every direction. To aim properly, the router and your phone exchange detailed measurements about how the waves are traveling through the room. That exchange is called beamforming feedback, and it's rich data about the physical space between the two devices. It also goes out with no encryption at all, which means a laptop with the right antenna parked outside the wall can capture it without ever joining the network.
The Karlsruhe team fed that feedback into a machine learning model and let it learn what each person's radio reflection looked like. After training, it could pick individuals out of the group whether they walked fast, walked slow, turned sideways, or wandered in from a different door. The signal reads your whole body at once, so your angle to the router barely matters the way it would for a camera.
The scientists spelled out the danger themselves. In their write-up they described how a city's existing WiFi could be used to follow people through a demonstration, with the routers already bolted to lamp posts, shops, cafés, and apartment walls quietly leaking beamforming data that a listener could stitch together into a trail. Nobody would need to install a single new device.
You know how to handle a camera. You can spot it, avoid it, cover it, pull up your hood. Radio waves give you nothing to react to, because they pass through your clothes and bounce off your skeleton whether you're looking at the router or not, and the room you're sitting in is full of them right now.
The researchers want the WiFi standard fixed so that beamforming feedback gets protected the way the rest of your traffic already is. Until that happens, the 197 people in that German lab are a preview of what any room can do. Their router learned all of them in a few hours, and none of them touched a keyboard, held a phone, or looked at a lens to make it happen.