The preprint argues that conventional biometric systems are increasingly unreliable in an era of AI-generated deepfakes, and proposes using the heart's electrical activity—ECG—as a biometric signal. The authors frame the ECG as a vibration signature, suggesting it may be harder to spoof than external traits like faces or voices.
The approach relies on TinyML, meaning the authentication model is designed to run on small, low-power devices rather than cloud servers. This could allow on-device verification, reducing privacy risks associated with transmitting biometric data.
The abstract does not provide experimental results or implementation details, so the practical accuracy and robustness of the method remain unspecified. As a single source, the claims are preliminary and have not been peer-reviewed.