AI powered radar can spy on phone calls from 10 feet, exposing new privacy risks

Penn State researchers have developed a system that uses millimeter-wave radar and AI to intercept phone conversations from up to 10 feet away, highlighting significant privacy risks. The technology captures subtle vibrations from a phone’s earpiece, decoding them into speech with approximately 60% accuracy. “When we talk on a cellphone, we tend to ignore the vibrations that come through the earpiece and cause the whole phone to vibrate,” said Suryoday Basak, a doctoral candidate in computer science.

By combining radar, used in self-driving cars and 5G networks, with an adapted AI speech recognition model called Whisper, the team transcribes conversations over a 10,000-word vocabulary. “If we capture these same vibrations using remote radars and bring in machine learning to help us learn what is being said, using context clues, we can determine whole conversations,” Basak explained.

Though imperfect, the system’s partial transcriptions could reveal sensitive information, similar to lip reading. The study aims to raise awareness about potential misuse by malicious actors, urging the development of protective measures to safeguard private communications.