The rise of AI has led to rapid technological progress, but also presents new challenges to personal privacy. Who-Fi technology exemplifies this concern, as it allows for the identification of individuals using Wi-Fi signals rather than cameras. This AI-driven system analyzes Wi-Fi signals to track activities and identify people, raising questions about digital privacy and security.
The technology functions by utilizing standard 2.4 GHz Wi-Fi signals to track individuals, as detailed in a research paper in arXiv. Who-Fi relies on a two-part system: Wi-Fi signals and transformer-based neural networks (large language models). These networks examine changes in Wi-Fi signals, known as Channel State Information, to understand how the signals behave in a room and how they are altered by a person’s presence. The system operates similarly to radar or sonar, detecting changes in signal strength and direction.
When a person is present, the Wi-Fi signal creates a unique pattern that can serve as a biometric identifier, comparable to fingerprints or facial recognition. Once trained, the system can track movement, re-identify individuals, and even capture data on body language, all without requiring cameras or microphones. The system’s design, using a single-antenna transmitter and a three-antenna receiver, makes it relatively inexpensive to deploy.