Researchers in Germany have unveiled a new artificial intelligence-based system capable of identifying individuals through ordinary WiFi signals with extremely high accuracy, sparking major concerns over privacy and invisible digital surveillance.
The study demonstrates how standard wireless networks can analyze radio wave movements to recognize people, even when they are not carrying smartphones or active connected devices.
According to the research team, the technology works by monitoring how radio waves travel and reflect within an environment, allowing AI systems to build detailed recognition patterns similar to how cameras process images.
WiFi Signals Used Like Invisible Cameras
Professor Thorsten Strufe from KASTEL, KIT’s Institute of Information Security and Dependability, explained that the system uses radio waves instead of light to identify individuals.
Researchers noted that turning off a smartphone may not prevent detection, since nearby devices connected to the same network continue generating enough wireless activity for the system to function effectively.
The technology reportedly allows AI to recognize individuals based on how radio signals interact with their movements, body shape, and presence within a room.
Ordinary Routers May Become Surveillance Tools
Experts involved in the project warned that the system could transform everyday WiFi routers into discreet monitoring devices capable of tracking people without attracting attention.
Researcher Julian Todt cautioned that individuals passing regularly through places such as cafés, offices, or airports could potentially be recognized automatically through nearby wireless networks.
The researchers say the biggest concern lies in the invisible nature of the technology, since wireless infrastructure already exists almost everywhere and generally does not raise suspicion among the public.
Felix Morsbach added that while intelligence agencies and cybercriminals already use other surveillance methods such as hacked cameras and smart devices, WiFi-based monitoring could eventually become far more widespread due to the global presence of wireless networks.
No Specialized Equipment Required
Unlike previous experimental systems that relied on expensive sensors or advanced laboratory equipment, the new method works using standard WiFi hardware already found in homes, businesses, and public spaces.
The technology takes advantage of communication data exchanged between routers and connected devices, specifically information known as beamforming feedback information (BFI). Since this data is often transmitted without encryption, researchers say it can potentially be intercepted and analyzed by nearby systems.
By processing these signal reflections, AI models can effectively generate multiple “views” of a person and learn to distinguish one individual from another.
Once trained, the system can reportedly identify a person within seconds.
Near-Perfect Accuracy Raises Privacy Fears
During testing involving 197 participants, researchers said the AI system achieved nearly 100% identification accuracy regardless of viewing angle or walking style.
The findings have intensified concerns over the future of privacy and civil liberties, particularly regarding how such technologies could be used by governments or organizations for covert monitoring.
Professor Strufe warned that despite the technology’s capabilities, it presents serious risks to fundamental privacy rights.
Researchers expressed particular concern over the possibility of authoritarian governments using the system to monitor protesters or track citizens without their knowledge.
Calls for Stronger Privacy Protections
The research team is now urging stronger safeguards and privacy protections to be incorporated into the upcoming IEEE 802.11bf WiFi standard in order to prevent potential misuse of the technology.
The project was funded under the Helmholtz “Engineering Secure Systems” initiative, and the findings are expected to be presented at the upcoming ACM Conference on Computer and Communications Security in Taipei.
