Test-Time Immunization: A Universal Defense Framework Against Jailbreaks for (Multimodal) Large Language Models

Yongcan Yu, Yanbo Wang, Ran He, Jian Liang

Abstract: While (multimodal) large language models (LLMs) have attracted widespread attention due to their exceptional capabilities, they remain vulnerable to jailbreak attacks. Various defense methods are proposed to defend against jailbreak attacks, however, they are often tailored to specific types of jailbreak attacks, limiting their effectiveness against diverse adversarial strategies. For instance, rephrasing-based defenses are effective against text adversarial jailbreaks but fail to counteract image-based attacks. To overcome these limitations, we propose a universal defense framework, termed Test-time IMmunization (TIM), which facilitates test-time optimization to counteract diverse jailbreak attacks. Specifically, TIM initially trains a gist token for efficient detection, which it subsequently applies to detect jailbreak activities during inference. When jailbreak attempts are identified, TIM implements safety fine-tuning using the detected jailbreak instructions paired with refusal answers. Furthermore, to mitigate potential performance degradation in the detector caused by parameter updates during safety fine-tuning, we decouple the fine-tuning process from the detection module. Extensive experiments on both LLMs and multimodal LLMs demonstrate the efficacy of TIM.

Primary Area: Social Aspects->Security

Keywords: Multimodal Large Lanugage Model, Jailbreak Defense, Test-Time Training, Large Language Model Security

Application-Driven Machine Learning: This submission is on Application-Driven Machine Learning.

Submission Number: 109