A Review of Eight Reasons Why Cybersecurity on Novel Generations of Brain-Computer Interfaces Must Be Prioritized [Póster]

Abstract

Brain-computer interfaces enable bidirectional communication between the brain and external devices, used in medicine for diagnosing and treating neurodegenerative diseases. Despite their advantages, they present vulnerabilities that attackers could exploit to cause brain damage. In this context, previous work defined three neural cyberattacks altering spontaneous neuronal activity. However, more effort is still needed to detect and characterize new neural cyberattacks with new behaviors. Based on that, this publication presents a taxonomy of eight neural cyberattacks, where five of them are novel. For each of them, this work offers a formal definition and the conceptualization of their behavior. Finally, it compares their impact on the short and long term, indicating that Neuronal Nonce was the most damaging attack in the short term, with an approximate 12% reduction of neural activity compared to spontaneous behavior. Finally, Neuronal Scanning was the most effective in the long term, offering a reduction of around 9%.