Play Guessing Game with LLM: Indirect Jailbreak Attack with Implicit Clues
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866913474868674560 |
|---|---|
| author | Chang, Zhiyuan Li, Mingyang Liu, Yi Wang, Junjie Wang, Qing Liu, Yang |
| author_facet | Chang, Zhiyuan Li, Mingyang Liu, Yi Wang, Junjie Wang, Qing Liu, Yang |
| contents | With the development of LLMs, the security threats of LLMs are getting more and more attention. Numerous jailbreak attacks have been proposed to assess the security defense of LLMs. Current jailbreak attacks primarily utilize scenario camouflage techniques. However their explicitly mention of malicious intent will be easily recognized and defended by LLMs. In this paper, we propose an indirect jailbreak attack approach, Puzzler, which can bypass the LLM's defense strategy and obtain malicious response by implicitly providing LLMs with some clues about the original malicious query. In addition, inspired by the wisdom of "When unable to attack, defend" from Sun Tzu's Art of War, we adopt a defensive stance to gather clues about the original malicious query through LLMs. Extensive experimental results show that Puzzler achieves a query success rate of 96.6% on closed-source LLMs, which is 57.9%-82.7% higher than baselines. Furthermore, when tested against the state-of-the-art jailbreak detection approaches, Puzzler proves to be more effective at evading detection compared to baselines. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_09091 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Play Guessing Game with LLM: Indirect Jailbreak Attack with Implicit Clues Chang, Zhiyuan Li, Mingyang Liu, Yi Wang, Junjie Wang, Qing Liu, Yang Cryptography and Security Artificial Intelligence Human-Computer Interaction With the development of LLMs, the security threats of LLMs are getting more and more attention. Numerous jailbreak attacks have been proposed to assess the security defense of LLMs. Current jailbreak attacks primarily utilize scenario camouflage techniques. However their explicitly mention of malicious intent will be easily recognized and defended by LLMs. In this paper, we propose an indirect jailbreak attack approach, Puzzler, which can bypass the LLM's defense strategy and obtain malicious response by implicitly providing LLMs with some clues about the original malicious query. In addition, inspired by the wisdom of "When unable to attack, defend" from Sun Tzu's Art of War, we adopt a defensive stance to gather clues about the original malicious query through LLMs. Extensive experimental results show that Puzzler achieves a query success rate of 96.6% on closed-source LLMs, which is 57.9%-82.7% higher than baselines. Furthermore, when tested against the state-of-the-art jailbreak detection approaches, Puzzler proves to be more effective at evading detection compared to baselines. |
| title | Play Guessing Game with LLM: Indirect Jailbreak Attack with Implicit Clues |
| topic | Cryptography and Security Artificial Intelligence Human-Computer Interaction |
| url | https://arxiv.org/abs/2402.09091 |