Advanced radar technology capable of intercepting nearby mobile phone conversations, presenting fresh privacy concerns.
In a groundbreaking development, a team of computer science researchers at Penn State University have created an experimental system that allows them to eavesdrop on phone calls remotely. This system, dubbed "wireless-tapping", uses millimeter-wave radar to detect and decode the tiny vibrations caused by speech on a phone’s earpiece, and then applies AI-based speech recognition to translate these vibration patterns into text transcripts.
The radar sensor, positioned about three meters (10 feet) away from the phone, captures the micro-vibrations generated by speech, which are normally imperceptible to human senses. These vibrations are then fed into advanced machine learning models, including adaptations of open-source AI speech recognition tools like Whisper, to decode these vibration patterns into recognizable human speech.
The AI training involves learning the relationships between vibration signatures and corresponding phonemes or words, allowing the reconstruction of spoken content with roughly 60% accuracy over a vocabulary of up to 10,000 words. Earlier versions of the technology decoded a smaller set of 10 keywords with higher accuracy (~83%) but less vocabulary coverage.
This combination of radar sensing and AI decoding allows the interception of phone calls without physical access or audio capture via conventional microphones, raising new privacy concerns as such non-invasive eavesdropping methods become technically feasible.
The study, published in the Proceedings of WiSec 2025: 18th ACM Conference on Security and Privacy in Wireless and Mobile Networks, serves as a crucial warning: even the faintest vibrations from your everyday devices can potentially betray your most private words.
The team stressed that their experiments are intended to highlight possible vulnerabilities before malicious actors exploit them. They also compared their approach to lip reading, which typically captures only 30% to 40% of spoken words but can still help people infer conversations when combined with context. However, the radar-AI system's output, though imperfect, can reveal sensitive information when supplemented with prior knowledge or manual correction.
The researchers envision future efforts to develop protective measures to secure personal conversations from this kind of remote surveillance. They also noted that even partial keyword matches could have serious security implications. This represents a novel and emerging surveillance threat leveraging the intersection of radar sensing and artificial intelligence.
- The experimental system developed by the computer science researchers at Penn State University, which allows for remote eavesdropping on phone calls, utilizes millimeter-wave radar technology and AI-based speech recognition to interpret the micro-vibrations caused by speech as text transcripts.
- The radar-AI system's output, though imperfect, can potentially reveal sensitive information when used in combination with prior knowledge or manual correction, as demonstrated by the team's experiments.
- As the combination of radar sensing and AI decoding allows for the interception of phone calls without physical access or conventional microphones, the researchers have stressed the importance of developing protective measures to secure personal conversations from remote surveillance, given the novel and emerging surveillance threat this represents.
