EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Red Teaming LLMs via Linguistic-Aware Fuzzing

Shuai Yuan, Nian Luo, Jingling Sun, Yihao Huang, Chengyu Zhang

Proceedings of the ACM on Software Engineering · 30.06.2026 · 3 (FSE) · 4482-4504

DOI: 10.1145/3808204 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the ACM on Software Engineering, online erschienen am 30.06.2026. Verfasst von Shuai Yuan und 4 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 38 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Safety alignment aims to prevent Large Language Models (LLMs) from producing harmful content. However, safety alignment remains vulnerable to malicious instructions. Red teaming is a critical methodology for identifying such vulnerabilities in LLMs. Existing approaches often rely on jailbreak templates or rule-based transformation, limiting the diversity of generated tests and the continued testing capability of these test approaches. To address these limitations, we propose Lingfuzz, a linguistic-aware fuzzing framework for continuing red teaming LLMs. The key idea of this framework is to mutate the existing malicious instructions at the lexical and syntactic levels, while keeping the malicious intentions of the instructions. Such mutations enable generating diverse malicious instructions due to the unlimited space of lexical and syntactic choices, while having a continued testing capability by iteratively mutating the mutants. We evaluated Lingfuzz on five aligned commercial LLMs and one open-source LLM in black-box testing. The results show that Lingfuzz triggers safety alignment vulnerabilities in 71.0% of the cases in JailbreakBench benchmark, higher than the second-highest baseline of 63.8%. Lingfuzz also possesses outstanding multilingual testing capabilities that far exceed other baselines with approximately 24% ESR advantage testing in Jailbench benchmark. The malicious instructions generated by Lingfuzz have almost three times higher diversity than previous work according to the self-BLEU metric. Lingfuzz also demonstrates strong continued testing capability by showing five times less sensitivity to the LLM evolution than other approaches.

Zitieren

APA

Yuan, S., Luo, N., Sun, J., Huang, Y., & Zhang, C. (2026). Red Teaming LLMs via Linguistic-Aware Fuzzing. Proceedings of the ACM on Software Engineering, 3(FSE), 4482-4504. https://doi.org/10.1145/3808204

Vancouver

Yuan S, Luo N, Sun J, Huang Y, Zhang C. Red Teaming LLMs via Linguistic-Aware Fuzzing. Proceedings of the ACM on Software Engineering. 2026;3(FSE):4482-4504. doi:10.1145/3808204

BibTeX
@article{Yuan2026,
  author = {Yuan, Shuai and Luo, Nian and Sun, Jingling and Huang, Yihao and Zhang, Chengyu},
  title = {Red Teaming LLMs via Linguistic-Aware Fuzzing},
  journal = {Proceedings of the ACM on Software Engineering},
  year = {2026},
  volume = {3},
  number = {FSE},
  pages = {4482--4504},
  doi = {10.1145/3808204}
}
Kostenlos starten →