Enhancing the Security & Privacy of Wearable Brain-Computer Interfaces / Jan 2022
Preprint on security vulnerabilities in brain-computer interfaces with Zahra Tarkhani, Lorena Qendro, and colleagues from Cambridge. This fascinating work analyzed security threats to wearable BCI devices from both an OS and adversarial ML perspective, discovering over 300 vulnerabilities across six attack vectors in real devices like Muse, NeuroSky, and OpenBCI. We introduced Argus, an information flow control system that mitigates these attacks with acceptable overhead - critical for protecting users' brainwave data and preventing remote attackers from compromising BCI-assisted devices.