@inproceedings{goel-etal-2025-turbofuzzllm,
    title = "{T}urbo{F}uzz{LLM}: Turbocharging Mutation-based Fuzzing for Effectively Jailbreaking Large Language Models in Practice",
    author = "Goel, Aman  and
      Wu, Xian  and
      Wang, Zhe  and
      Bespalov, Dmitriy  and
      Qi, Yanjun",
    editor = "Chen, Weizhu  and
      Yang, Yi  and
      Kachuee, Mohammad  and
      Fu, Xue-Yong",
    booktitle = "Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 3: Industry Track)",
    month = apr,
    year = "2025",
    address = "Albuquerque, New Mexico",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/landing_page/2025.naacl-industry.43/",
    pages = "523--534",
    ISBN = "979-8-89176-194-0",
    abstract = "Jailbreaking large-language models (LLMs) involves testing their robustness against adversarial prompts and evaluating their ability to withstand prompt attacks that could elicit unauthorized or malicious responses. In this paper, we present TurboFuzzLLM, a mutation-based fuzzing technique for efficiently finding a collection of effective jailbreaking templates that, when combined with harmful questions, can lead a target LLM to produce harmful responses through black-box access via user prompts. We describe the limitations of directly applying existing template-based attacking techniques in practice, and present functional and efficiency-focused upgrades we added to mutation-based fuzzing to generate effective jailbreaking templates automatically. TurboFuzzLLM achieves $\geq$ 95{\%} attack success rates (ASR) on public datasets for leading LLMs (including GPT-4o {\&} GPT-4 Turbo), shows impressive generalizability to unseen harmful questions, and helps in improving model defenses to prompt attacks."
}