EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC 4.0

Research on Information Security Vulnerability Testing and Verification Methods and Protection Countermeasures for LLM Applications in Intelligent Cockpits

Ziyi Wang, Yihong Qin, Benyin Wu, Xuesong Wu

Journal of Computer Science and Artificial Intelligence · 30.04.2026 · 7 (1) · 52-55

DOI: 10.54097/qnhjph62 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Computer Science and Artificial Intelligence, online erschienen am 30.04.2026. Verfasst von Ziyi Wang und 3 weiteren Autor:innen. Die Arbeit nennt 5 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In recent years, the application of large language models (LLMs) in intelligent cockpits has been continuously deepened, significantly improving the natural human-vehicle interaction experience. However, the generative characteristics of models, multi-modal data interaction, and access to vehicle control privileges have also introduced a series of new information security vulnerabilities such as prompt injection, data privacy leakage, unauthorized vehicle manipulation, and model hallucinations. Traditional in-vehicle security testing systems mostly target vulnerabilities in in-vehicle operating systems, in-vehicle buses, and network communications, and are difficult to adapt to the semantic-level and generative security risks brought by LLMs. To address this issue, this paper classifies and analyzes the security vulnerabilities of LLM applications in intelligent cockpit scenarios, constructs a vulnerability testing and verification method integrating static auditing, dynamic fuzz testing, red team adversarial testing, and multi-modal boundary testing, and establishes a corresponding security evaluation index system. On this basis, a hierarchical protection strategy is proposed from five dimensions: input protection, model hardening, system isolation, data security, and compliance auditing. Verified through real-vehicle environment testing, the proposed testing method can effectively identify 92.7% of known security vulnerabilities. After deploying the protection countermeasures, the relevant attack success rate drops from 38.5% to 4.3%, which can provide technical support for the secure development, testing verification, and engineering implementation of LLMs in intelligent cockpits.

Zitieren

APA

Wang, Z., Qin, Y., Wu, B., & Wu, X. (2026). Research on Information Security Vulnerability Testing and Verification Methods and Protection Countermeasures for LLM Applications in Intelligent Cockpits. Journal of Computer Science and Artificial Intelligence, 7(1), 52-55. https://doi.org/10.54097/qnhjph62

Vancouver

Wang Z, Qin Y, Wu B, Wu X. Research on Information Security Vulnerability Testing and Verification Methods and Protection Countermeasures for LLM Applications in Intelligent Cockpits. Journal of Computer Science and Artificial Intelligence. 2026;7(1):52-55. doi:10.54097/qnhjph62

BibTeX
@article{Wang2026,
  author = {Wang, Ziyi and Qin, Yihong and Wu, Benyin and Wu, Xuesong},
  title = {Research on Information Security Vulnerability Testing and Verification Methods and Protection Countermeasures for LLM Applications in Intelligent Cockpits},
  journal = {Journal of Computer Science and Artificial Intelligence},
  year = {2026},
  volume = {7},
  number = {1},
  pages = {52--55},
  doi = {10.54097/qnhjph62}
}
Kostenlos starten →