Advance AI technology enables radar to intercept mobile conversations from mere feet away, raising fresh concerns over privacy and potential information breaches.
A team of computer science researchers at Penn State have made a groundbreaking discovery, developing a technology that combines millimeter-wave radar and artificial intelligence (AI) to potentially eavesdrop on phone calls remotely. The research, supported by the U.S. National Science Foundation, has raised significant concerns about privacy as the technology could intercept phone calls by decoding tiny vibrations from a phone's earpiece.
Millimeter-Wave Radar Technology
The technology works by detecting the vibrations caused by a person speaking on a phone. These subtle vibrations are captured using millimeter-wave radar sensors, which are highly sensitive and can detect them from a distance of up to 10 feet (about 3 meters).
Artificial Intelligence (AI) Integration
The radar sensor collects data on the vibrations, which are then processed by AI algorithms. To interpret the noisy and low-quality data, the researchers adapted Whisper, an open-source AI speech recognition model, using a low-rank adaptation machine learning technique. This allowed them to retrain just 1 percent of Whisper's parameters specifically for radar data, improving transcription results without rebuilding the entire model from scratch.
Transcription and Accuracy
By combining the radar data with AI, the researchers can transcribe conversations with approximately 60% accuracy over a vocabulary of up to 10,000 words. While this accuracy is not perfect, the technology raises significant privacy concerns as even partial data can reveal sensitive information when combined with contextual clues.
The output of the model, though imperfect, can reveal sensitive information when supplemented with prior knowledge or manual correction. Partial keyword matches from the radar-AI system's output can have serious security implications. The output of the model combined with contextual information can allow the inference of parts of a phone conversation from a few meters away.
Privacy Concerns and Future Efforts
The research raises significant privacy concerns about the potential misuse of such emerging technologies. The team envisioned future efforts to develop protective measures to secure personal conversations from this kind of remote surveillance.
The study has been published in the Proceedings of WiSec 2025: 18th ACM Conference on Security and Privacy in Wireless and Mobile Networks.
The technology's ability to transcribe conversations, although not perfect, can still reveal sensitive information when supplemented with prior knowledge or manual correction. This raises concerns about cybersecurity, as partial keyword matches from the radar-AI system's output can have serious implications for privacy. Moreover, the potential misuse of such advanced technologies necessitates future innovation in developing protective measures to secure personal conversations from remote surveillance.
