Gespeichert in:
| Hauptverfasser: | , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | https://arxiv.org/abs/2503.08195 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866916649578266624 |
|---|---|
| author | Meng, Wenlong Zhang, Fan Yao, Wendao Guo, Zhenyuan Li, Yuwei Wei, Chengkun Chen, Wenzhi |
| author_facet | Meng, Wenlong Zhang, Fan Yao, Wendao Guo, Zhenyuan Li, Yuwei Wei, Chengkun Chen, Wenzhi |
| contents | Large language models (LLMs) have demonstrated significant utility in a wide range of applications; however, their deployment is plagued by security vulnerabilities, notably jailbreak attacks. These attacks manipulate LLMs to generate harmful or unethical content by crafting adversarial prompts. While much of the current research on jailbreak attacks has focused on single-turn interactions, it has largely overlooked the impact of historical dialogues on model behavior. In this paper, we introduce a novel jailbreak paradigm, Dialogue Injection Attack (DIA), which leverages the dialogue history to enhance the success rates of such attacks. DIA operates in a black-box setting, requiring only access to the chat API or knowledge of the LLM's chat template. We propose two methods for constructing adversarial historical dialogues: one adapts gray-box prefilling attacks, and the other exploits deferred responses. Our experiments show that DIA achieves state-of-the-art attack success rates on recent LLMs, including Llama-3.1 and GPT-4o. Additionally, we demonstrate that DIA can bypass 5 different defense mechanisms, highlighting its robustness and effectiveness. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_08195 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Dialogue Injection Attack: Jailbreaking LLMs through Context Manipulation Meng, Wenlong Zhang, Fan Yao, Wendao Guo, Zhenyuan Li, Yuwei Wei, Chengkun Chen, Wenzhi Computation and Language Large language models (LLMs) have demonstrated significant utility in a wide range of applications; however, their deployment is plagued by security vulnerabilities, notably jailbreak attacks. These attacks manipulate LLMs to generate harmful or unethical content by crafting adversarial prompts. While much of the current research on jailbreak attacks has focused on single-turn interactions, it has largely overlooked the impact of historical dialogues on model behavior. In this paper, we introduce a novel jailbreak paradigm, Dialogue Injection Attack (DIA), which leverages the dialogue history to enhance the success rates of such attacks. DIA operates in a black-box setting, requiring only access to the chat API or knowledge of the LLM's chat template. We propose two methods for constructing adversarial historical dialogues: one adapts gray-box prefilling attacks, and the other exploits deferred responses. Our experiments show that DIA achieves state-of-the-art attack success rates on recent LLMs, including Llama-3.1 and GPT-4o. Additionally, we demonstrate that DIA can bypass 5 different defense mechanisms, highlighting its robustness and effectiveness. |
| title | Dialogue Injection Attack: Jailbreaking LLMs through Context Manipulation |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2503.08195 |