AI Security Technology Developed to Detect Autonomous Vehicle Hacks With 99.87% Accuracy
- Input
- 2026-11-16 20:24:23
- Updated
- 2026-11-16 20:24:23

[Financial News] A security technology has been developed that detects external hacking attacks targeting autonomous or connected cars in motion with 99.87% accuracy and explains in detail why the artificial intelligence (AI) classified them as hacks. Sookmyung Women's University announced on the 16th that a research team comprising Professor Jung Seong-hun of the Department of AI Engineering and undergraduate student Kim Ji-su, who majors in data science, had developed an AI technology called “X-SPUR” that detects cyberattacks on in-vehicle communication networks and explains the basis for its decisions.
The research was led by undergraduate student Kim Ji-su as first author, earning recognition when the paper was published in a world-renowned international journal. The findings were published last month in IEEE Transactions on Industrial Informatics, an international journal ranked in the top 8.5% of the field of industrial informatics.
As cars have increasingly become connected to communication networks, the volume of data exchanged among electronic devices inside vehicles has surged, along with the risk of hacking. However, it is difficult to obtain enough data on real-world vehicle hacking incidents. Existing AI security technologies also have a limitation: even when they detect an anomaly, they cannot explain exactly which communication signal is problematic, making it difficult for security personnel to respond immediately.
To address this problem, the research team chose to train the AI only on vehicles' normal communication data, without using hacking data. When a signal falls outside the normal range, the system immediately classifies it as an attack and shows in detail which information among the components of the communication signal affected its anomaly assessment. This enables security personnel to see the basis for the AI's decision for themselves, analyze the cause of the hack, and respond quickly.
In tests using representative in-vehicle communication data, the technology recorded a high detection rate of 99.87%. It is expected to play a key role in improving driving safety and security reliability for connected and autonomous vehicles.
Meanwhile, Professor Jung and student Kim's research team also achieved results in security research aimed at preventing system overload by analyzing access traffic to e-commerce websites, including online shopping malls. The team developed a technology that analyzes approximately 12.7 million real shopping-mall access records and prioritizes the detection of malicious or overload-inducing connections that can damage entire systems. The paper was also selected for presentation at the IEEE BigData 2026 conference, scheduled to take place next month in Phoenix, United States.
[email protected] Kim Man-gi Reporter