［1］Xu Z H, Liu Y, Deng G L, et al. LLM jailbreak attack versus defense techniquesa comprehensive study［EBOL］. 2024 ［20250816］. https:arxiv.orgpdf2402.13457v1［2］李南, 丁益东, 江浩宇, 等. 面向大语言模型的越狱攻击综述［J］. 计算机研究与发展, 2024, 61(5): 11561181［3］Li X R, Wang R C, Cheng M H, et al. Drattack: Prompt decomposition and reconstruction makes powerful llm jailbreakers ［EBOL］. 2024 ［20250816］. https:arxiv.orgpdf2402.16914［4］Chang Z Y, Li M Y, Liu Y, et al. Play guessing game with llm: Indirect jailbreak attack with implicit clues［EBOL］. 2024 ［20250816］. https:arxiv.orgpdf2402.09091［5］Zhang T R, Cao B C, Cao Y P, et al. Wordgame: Efficient & effective LLM jailbreak via simultaneous obfuscation in query and response［EBOL］. 2024 ［20250816］. https:arxiv.orgpdf2405.14023［6］Zhou Y K, Huang Z J, Lu F Y, et al. Don’t say no: Jailbreaking LLM by suppressing refusal［EBOL］. 2024 ［20250816］. https:arxiv.orgpdf2404.16369［7］Chao P, Robey A, Dobriban E, et al. Jailbreaking black box largelanguagemodels in twenty queries［EBOL］. 2023 ［20250816］. https:arxiv.orgpdf2310.08419［8］Yu, JH, Lin X W, Yu Z, et al. Gptfuzzer: Red teaming large language models with autogenerated jailbreak prompts［EBOL］. 2023 ［20250816］. https:arxiv.orgpdf2309.10253v3［9］Zou A, Wang Z F, Carlini N, et al. Universal and transferable adversarial attacks on aligned language models［EBOL］. 2023 ［20250816］. https:www.cnblogs.comLittleHannp17590737.html［10］Wu T Y,Xue Z W,Liu Y, et al. Geneshift: Impact of different scenario shift on Jailbreaking LLM［EBOL］. 2025 ［20250816］. https:arxiv.orgpdf2504.08104［11］Speer R, Chin J, Havasi C. ConceptNet 5.5: An open multilingual graph of general knowledge［C］ Proc of the 31st AAAI Conf on Artificial Intelligence. New York: ACM, 2017: 44444451［12］Schulman J, Wolski F, Dhariwal P, et al. Proximal policy optimization algorithms［EBOL］. 2017 ［20250816］. https:arxiv.orgpdf1707.06347 |
