Ordinary WiFi Can Now Identify People With Near-Perfect Accuracy

WiFi used to be the invisible plumbing of modern life, quietly pushing data from router to screen. Research covered by ScienceDaily on 5/22/2026 says that same everyday signal can be turned into a near-perfect ID check, recognizing people by reading how radio waves bounce around a room. The work, from Karlsruhe Institute of Technology’s KASTEL Institute of Information Security and Dependability, uses standard off-the-shelf routers as passive sensors, pulling a signature from reflection patterns even if you are not carrying a device and your phone is off. In tests summarized by multiple outlets, the system dubbed BFId reportedly hit about 99.5% accuracy across 197 subjects, adding fresh urgency to the privacy question of what ordinary networks in homes and public spaces could quietly reveal.

WiFi: more than just internet access

If you’re like me, you probably think of WiFi as the quiet utility in the background: a router on a shelf, a password on a sticky note, and the occasional dead zone. But a May 2026 research write-up made WiFi feel a lot less invisible. In a May 22, 2026 story, ScienceDaily described how scientists in Germany identified people using only everyday wireless signals.

The technique: turning reflections into identity

The work came from the Karlsruhe Institute of Technology (KIT), specifically its KASTEL Institute of Information Security and Dependability. The idea is straightforward but unsettling: standard, off-the-shelf WiFi routers can act like a passive sensing system, because radio waves bounce off bodies and objects and leave behind reflection patterns that can be converted into a signature.

Coverage describes the system as BFId, built around WiFi beamforming feedback data. Reportedly, it hit 99.5% accuracy when identifying people across 197 subjects. Importantly, it does not depend on someone carrying a device, and it can still work even if a phone is turned off.

When a home network starts to look like surveillance

The researchers themselves emphasized how reachable this is with common gear: “We have shown robust identity inference with common-of-the-shelf hardware which is already in widespread adoption in many homes and public areas.” That detail matters for Americans because “specialized equipment” is usually where cost and deployment friction show up. Here, the hardware is already in apartments, offices, coffee shops, and campuses.

It also goes beyond naming a person. The team showed WiFi signals can reveal whether people are present, where they are located, and what they are doing, a point echoed by coverage describing activity detection in addition to identification, such as in phone-off detection. How should consumer privacy expectations change if ordinary networking can double as a sensor?

A line from research into the real world

This isn’t the first time researchers have tried to identify humans through WiFi. An earlier IEEE conference paper titled “WiFi-ID: Human Identification Using WiFi Signal” explored identification using WiFi signals and gait patterns years before the 2026 KIT work.

What feels different now is the combination of specificity and accessibility: a named system, high reported accuracy, and an explicit acknowledgement that the necessary hardware is already widespread. For US tech companies building smart-home and enterprise networking products, and for regulators who think about covert monitoring, that’s a new set of assumptions to confront.