Akademik Veri Yönetim Sistemi
The 27th Information Security Conference (ISC 2024), 23 - 25 Ekim 2024, ss.183-202
Brain-Computer Interface (BCI) has become an exciting area in the gaming and entertainment industry, which reads mind silently to capture inputs and interacts with the computing system without user intervention to provide a more user-friendly experience. Such silent mind-reading technology can bring serious privacy threats to end-users if proper precautions are not implemented before massive industrialization. In this paper, we present a novel attack, News Reader, which infers brainwaves silently when people are reading news articles on the Internet. In particular, we investigate neural patterns related to different news categories to identify what type of news the victim is reading by utilizing a single-electrode and low-cost consumer-grade EEG headset.
The result confirms that significant differences are present when human brain processes different types of news articles (i.e., Political News, Daily News, and Others News). We evaluated the effectiveness of News Reader attack through statistical analysis, time-frequency analysis, and machine learning. All methodologies could successfully distinguish news categories from brainwaves and our machine learning achieved accuracy of up to 99.79% in correctly classifying news categories from neural pattern. The scientific evaluation of the News Reader attack serves as a fundamental research direction for security researchers to add security layers to BCI technology, especially to consumer-grade EEG technology before our private sensitive information gets exposed to the malicious parties. We believe, our work provides a notable insight for researchers of current and future BCI technology and raises the awareness of unprotected use of brainwaves can lead to the most serious privacy attacks in human life.