AI is pushing technological boundaries, but at the same time, it’s amplifying privacy risks. Who-Fi is a prime example of this: a technology that can identify individuals without using cameras. This system, powered by artificial intelligence, tracks people’s activities by analyzing Wi-Fi signals.
The core of Who-Fi’s operation lies in its use of standard 2.4 GHz Wi-Fi signals, detailed in a research paper published in arXiv. The system uses two components: Wi-Fi signals and transformer-based neural networks (large language models). It reads changes in the Wi-Fi signals, known as channel state information, to understand how they behave in a space and how a person’s presence alters them, similar to how radar or sonar works.
The interaction between a person and a Wi-Fi signal creates a unique pattern that can be used for biometric identification. After training, the system can track activities, re-identify individuals entering the network, and even recognize sign language. The system’s key advantage is its ability to operate without cameras or microphones. The research indicates that the system requires a single-antenna transmitter and a three-antenna receiver, making it an affordable surveillance solution.