AI-Powered Radar Enables Remote Eavesdropping on Phone Calls
Researchers at Pennsylvania State University developed an AI-powered millimeter-wave radar system that can reportedly intercept phone calls from three meters away by detecting minute vibrations from the phone speaker, achieving approximately 60 percent transcription accuracy. The researchers note the technology raises privacy concerns and could have security implications even when word recognition is incomplete.
What happened
Researchers at Pennsylvania State University developed a method to remotely intercept phone calls by detecting minute vibrations emitted from a phone's speaker using millimeter-wave radar combined with an AI speech recognition system.
The system can record and transcribe conversations from a distance of approximately three meters with approximately 60 percent accuracy by decoding subtle vibrations from the phone speaker. The team used the open-source AI tool Whisper with minimal retraining on radar data to process the noisy radar information.
Reported impact
The technology raises significant privacy concerns and the potential for misuse of such emerging surveillance capabilities.
Index analysis
The research builds on a 2022 project in which the same team achieved 83 percent accuracy in identifying 10 predefined words using a similar approach. The new work extends this capability to continuous speech transcription, though accuracy is lower due to the complexity of decoding noisy radar data.
Researchers warned that even incomplete recognition of key words could have significant security implications.
Classification
Sources and evidence
Sources not yet published
This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.