CoChat
Sign Up
Cited by 24Adversarial Robustness in Machine LearningSecurity and Verification in ComputingTopic Modeling

Improved Few-Shot Jailbreaking Can Circumvent Aligned Language Models and Their Defenses

Open access
Zheng, Xiaosen, Pang, Tianyu, Du, Chao, Liu, Qian, Jiang, Jing, Lin, Min·2024·Advances in Neural Information Processing Systems 37
DOI 10.52202/079017-1034
Open this paper in CoChat — free accountOpen PDF

Abstract

Recently, Anil et al. (2024) show that many-shot (up to hundreds of) demonstrations can jailbreak state-of-the-art LLMs by exploiting their long-context capability. Nevertheless, is it possible to use few-shot demonstrations to efficiently jailbreak LLMs within limited context sizes? While the vanilla few-shot jailbreaking may be inefficient, we propose improved techniques such as injecting special system tokens like [/INST] and employing demo-level random search from a collected demo pool. These simple techniques result in surprisingly effective jailbreaking against aligned LLMs (even with advanced defenses). For examples, our method achieves >80% (mostly >95%) ASRs on Llama-2-7B and Llama-3-8B without multiple restarts, even if the models are enhanced by strong defenses such as perplexity detection and/or SmoothLLM, which is challenging for suffix-based jailbreaking. In addition, we conduct comprehensive and elaborate (e.g., making sure to use correct system prompts) evaluations against other aligned LLMs and advanced defenses, where our method consistently achieves nearly 100% ASRs. Our code is available at https://github.com/sail-sg/I-FSJ.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.52202/079017-1034

Cites

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • In-context learning can enable overcoming safety fine-tuning.

    “or the ability to overcome safety fine-tuning (Anil et al., 2024;Zheng et al., 2024).”

    From Machine Psychology (2023) · cites this paper as Zheng et al., 2024

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

No references could be resolved yet.

Research works printed but not linked: 69

  1. Gabriel Alon and Michael Kamfonas. Detecting language model attacks with perplexity. arXiv preprint arXiv:2308.14132, 2023. · arXiv
  2. Maksym Andriushchenko, Francesco Croce, and Nicolas Flammarion. Jailbreaking leading safety-aligned llms with simple adaptive attacks. arXiv preprint arXiv:2404.02151, 2024. · arXiv
  3. Jinze Bai, Shuai Bai, Yunfei Chu, Zeyu Cui, Kai Dang, Xiaodong Deng, Yang Fan, Wenbin Ge, Yu Han, Fei Huang, et al. Qwen technical report. arXiv preprint arXiv:2309.16609, 2023. · arXiv
  4. Yuntao Bai, Andy Jones, Kamal Ndousse, Amanda Askell, Anna Chen, Nova DasSarma, Dawn Drain, Stanislav Fort, Deep Ganguli, Tom Henighan, et al. Training a helpful and harmless assistant with reinforcement learning from human feedback. arXiv preprint arXiv:2204.05862, 2022. · arXiv
  5. Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. Language models are few-shot learners. In Advances in Neural Information Processing Systems (NeurIPS), 2020.
  6. Bochuan Cao, Yuanpu Cao, Lu Lin, and Jinghui Chen. Defending against alignment-breaking attacks via robustly aligned llm. arXiv preprint arXiv:2309.14348, 2023. · arXiv
  7. Nicholas Carlini, Milad Nasr, Christopher A Choquette-Choo, Matthew Jagielski, Irena Gao, Pang Wei Koh, Daphne Ippolito, Florian Tramèr, and Ludwig Schmidt. Are aligned neural networks adversarially aligned? In Advances in Neural Information Processing Systems (NeurIPS), 2023.
  8. Patrick Chao, Alexander Robey, Edgar Dobriban, Hamed Hassani, George J Pappas, and Eric Wong. Jailbreaking black box large language models in twenty queries. arXiv preprint arXiv:2310.08419, 2023. · arXiv
  9. Patrick Chao, Edoardo Debenedetti, Alexander Robey, Maksym Andriushchenko, Francesco Croce, Vikash Sehwag, Edgar Dobriban, Nicolas Flammarion, George J Pappas, Florian Tramer, et al. Jailbreakbench: An open robustness benchmark for jailbreaking large language models. arXiv preprint arXiv:2404.01318, 2024. · arXiv
  10. Gelei Deng, Yi Liu, Yuekang Li, Kailong Wang, Ying Zhang, Zefeng Li, Haoyu Wang, Tianwei Zhang, and Yang Liu. Jailbreaker: Automated jailbreak across multiple large language model chatbots. arXiv preprint arXiv:2307.08715, 2023. · arXiv
  11. Yue Deng, Wenxuan Zhang, Sinno Jialin Pan, and Lidong Bing. Multilingual jailbreak chal- lenges in large language models. arXiv preprint arXiv:2310.06474, 2023. · arXiv
  12. Pranav Gade, Simon Lermen, Charlie Rogers-Smith, and Jeffrey Ladish. Badllama: cheaply removing safety fine-tuning from llama 2-chat 13b. arXiv preprint arXiv:2311.00117, 2023. · arXiv
  13. Deep Ganguli, Liane Lovitt, Jackson Kernion, Amanda Askell, Yuntao Bai, Saurav Kadavath, Ben Mann, Ethan Perez, Nicholas Schiefer, Kamal Ndousse, et al. Red teaming language models to reduce harms: Methods, scaling behaviors, and lessons learned. arXiv preprint arXiv:2209.07858, 2022. · arXiv
  14. Kai Greshake, Sahar Abdelnabi, Shailesh Mishra, Christoph Endres, Thorsten Holz, and Mario Fritz. Not what you've signed up for: Compromising real-world llm-integrated applications with indirect prompt injection. In ACM Workshop on Artificial Intelligence and Security, 2023.
  15. Jonathan Hayase, Ema Borevkovic, Nicholas Carlini, Florian Tramèr, and Milad Nasr. Query- based adversarial prompt generation. arXiv preprint arXiv:2402.12329, 2024. · arXiv
  16. Alec Helbling, Mansi Phute, Matthew Hull, and Duen Horng Chau. Llm self defense: By self examination, llms know they are being tricked. arXiv preprint arXiv:2308.07308, 2023. · arXiv
  17. Xiaomeng Hu, Pin-Yu Chen, and Tsung-Yi Ho. Gradient cuff: Detecting jailbreak attacks on large language models by exploring refusal loss landscapes. arXiv preprint arXiv:2403.00867, 2024. · arXiv
  18. Zhengmian Hu, Gang Wu, Saayan Mitra, Ruiyi Zhang, Tong Sun, Heng Huang, and Vishy Swaminathan. Token-level adversarial prompt detection based on perplexity measures and contextual information. arXiv preprint arXiv:2311.11509, 2023. · arXiv
  19. Yangsibo Huang, Samyak Gupta, Mengzhou Xia, Kai Li, and Danqi Chen. Catastrophic jailbreak of open-source llms via exploiting generation. In International Conference on Learning Representations (ICLR), 2024.
  20. Hakan Inan, Kartikeya Upasani, Jianfeng Chi, Rashi Rungta, Krithika Iyer, Yuning Mao, Michael Tontchev, Qing Hu, Brian Fuller, Davide Testuggine, et al. Llama guard: Llm-based input-output safeguard for human-ai conversations. arXiv preprint arXiv:2312.06674, 2023. · arXiv
  21. Neel Jain, Avi Schwarzschild, Yuxin Wen, Gowthami Somepalli, John Kirchenbauer, Ping-yeh Chiang, Micah Goldblum, Aniruddha Saha, Jonas Geiping, and Tom Goldstein. Baseline de- fenses for adversarial attacks against aligned language models. arXiv preprint arXiv:2309.00614, 2023. · arXiv
  22. Jiabao Ji, Bairu Hou, Alexander Robey, George J Pappas, Hamed Hassani, Yang Zhang, Eric Wong, and Shiyu Chang. Defending large language models against jailbreak attacks via semantic smoothing. arXiv preprint arXiv:2402.16192, 2024. · arXiv
  23. Albert Q Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lucile Saulnier, et al. Mistral 7b. arXiv preprint arXiv:2310.06825, 2023. · arXiv
  24. Tomasz Korbak, Kejian Shi, Angelica Chen, Rasika Vinayak Bhalerao, Christopher Buckley, Jason Phang, Samuel R Bowman, and Ethan Perez. Pretraining language models with human preferences. In International Conference on Machine Learning (ICML), 2023.
  25. Aounon Kumar, Chirag Agarwal, Suraj Srinivas, Soheil Feizi, and Hima Lakkaraju. Certifying llm safety against adversarial prompting. arXiv preprint arXiv:2309.02705, 2023. · arXiv
  26. Raz Lapid, Ron Langberg, and Moshe Sipper. Open sesame! universal black box jailbreaking of large language models. arXiv preprint arXiv:2309.01446, 2023. · arXiv
  27. Simon Lermen, Charlie Rogers-Smith, and Jeffrey Ladish. Lora fine-tuning efficiently undoes safety training in llama 2-chat 70b. arXiv preprint arXiv:2310.20624, 2023. · arXiv
  28. Xuan Li, Zhanke Zhou, Jianing Zhu, Jiangchao Yao, Tongliang Liu, and Bo Han. Deepinception: Hypnotize large language model to be jailbreaker. arXiv preprint arXiv:2311.03191, 2023. · arXiv
  29. Yuhui Li, Fangyun Wei, Jinjing Zhao, Chao Zhang, and Hongyang Zhang. Rain: Your language models can align themselves without finetuning. arXiv preprint arXiv:2309.07124, 2023. · arXiv
  30. Zeyi Liao and Huan Sun. Amplegcg: Learning a universal and transferable generative model of adversarial suffixes for jailbreaking both open and closed llms. arXiv preprint arXiv:2404.07921, 2024. · arXiv
  31. Bill Yuchen Lin, Abhilasha Ravichander, Ximing Lu, Nouha Dziri, Melanie Sclar, Khyathi Chandu, Chandra Bhagavatula, and Yejin Choi. The unlocking spell on base llms: Rethinking alignment via in-context learning. arXiv preprint arXiv:2312.01552, 2023. · arXiv
  32. Nelson F Liu, Kevin Lin, John Hewitt, Ashwin Paranjape, Michele Bevilacqua, Fabio Petroni, and Percy Liang. Lost in the middle: How language models use long contexts. Transactions of the Association for Computational Linguistics (TACL), 2023.
  33. Xiaogeng Liu, Nan Xu, Muhao Chen, and Chaowei Xiao. Autodan: Generating stealthy jailbreak prompts on aligned large language models. arXiv preprint arXiv:2310.04451, 2023. · arXiv
  34. Yi Liu, Gelei Deng, Zhengzi Xu, Yuekang Li, Yaowen Zheng, Ying Zhang, Lida Zhao, Tianwei Zhang, and Yang Liu. Jailbreaking chatgpt via prompt engineering: An empirical study. arXiv preprint arXiv:2305.13860, 2023. · arXiv
  35. Neal Mangaokar, Ashish Hooda, Jihye Choi, Shreyas Chandrashekaran, Kassem Fawaz, Somesh Jha, and Atul Prakash. Prp: Propagating universal perturbations to attack large language model guard-rails. arXiv preprint arXiv:2402.15911, 2024. · arXiv
  36. Mantas Mazeika, Long Phan, Xuwang Yin, Andy Zou, Zifan Wang, Norman Mu, Elham Sakhaee, Nathaniel Li, Steven Basart, Bo Li, et al. Harmbench: A standardized evaluation framework for automated red teaming and robust refusal. arXiv preprint arXiv:2402.04249, 2024. · arXiv
  37. Anay Mehrotra, Manolis Zampetakis, Paul Kassianik, Blaine Nelson, Hyrum Anderson, Yaron Singer, and Amin Karbasi. Tree of attacks: Jailbreaking black-box llms automatically. arXiv preprint arXiv:2312.02119, 2023. · arXiv
  38. Yichuan Mo, Yuji Wang, Zeming Wei, and Yisen Wang. Studious bob fight back against jailbreaking via prompt adversarial tuning. arXiv preprint arXiv:2402.06255, 2024. · arXiv
  39. Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. Training language models to follow instructions with human feedback. In Advances in Neural Information Processing Systems (NeurIPS), 2022.
  40. Anselm Paulus, Arman Zharmagambetov, Chuan Guo, Brandon Amos, and Yuandong Tian. Advprompter: Fast adaptive adversarial prompting for llms. arXiv preprint arXiv:2404.16873, 2024. · arXiv
  41. Ethan Perez, Saffron Huang, Francis Song, Trevor Cai, Roman Ring, John Aslanides, Amelia Glaese, Nat McAleese, and Geoffrey Irving. Red teaming language models with language models. arXiv preprint arXiv:2202.03286, 2022. · arXiv
  42. Xiangyu Qi, Yi Zeng, Tinghao Xie, Pin-Yu Chen, Ruoxi Jia, Prateek Mittal, and Peter Henderson. Fine-tuning aligned language models compromises safety, even when users do not intend to! arXiv preprint arXiv:2310.03693, 2023. · arXiv
  43. Abhinav Rao, Sachin Vashistha, Atharva Naik, Somak Aditya, and Monojit Choudhury. Tricking llms into disobedience: Understanding, analyzing, and preventing jailbreaks. arXiv preprint arXiv:2305.14965, 2023. · arXiv
  44. N Reimers. Sentence-bert: Sentence embeddings using siamese bert-networks. arXiv preprint arXiv:1908.10084, 2019. · arXiv
  45. Alexander Robey, Eric Wong, Hamed Hassani, and George J Pappas. Smoothllm: Defending large language models against jailbreaking attacks. arXiv preprint arXiv:2310.03684, 2023. · arXiv
  46. Yangjun Ruan, Honghua Dong, Andrew Wang, Silviu Pitis, Yongchao Zhou, Jimmy Ba, Yann Dubois, Chris J Maddison, and Tatsunori Hashimoto. Identifying the risks of lm agents with an lm-emulated sandbox. arXiv preprint arXiv:2309.15817, 2023. · arXiv
  47. Reshabh K Sharma, Vinayak Gupta, and Dan Grossman. Spml: A dsl for defending language models against prompt attacks. arXiv preprint arXiv:2402.11755, 2024. · arXiv
  48. Xinyue Shen, Zeyuan Chen, Michael Backes, Yun Shen, and Yang Zhang. "do anything now": Characterizing and evaluating in-the-wild jailbreak prompts on large language models. arXiv preprint arXiv:2308.03825, 2023. · arXiv
  49. Chawin Sitawarin, Norman Mu, David Wagner, and Alexandre Araujo. Pal: Proxy-guided black-box attack on large language models. arXiv preprint arXiv:2402.09674, 2024. · arXiv
  50. Yu Tian, Xiao Yang, Jingyuan Zhang, Yinpeng Dong, and Hang Su. Evil geniuses: Delving into the safety of llm-based agents. arXiv preprint arXiv:2311.11855, 2023. · arXiv
  51. Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, et al. Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288, 2023. · arXiv
  52. Sam Toyer, Olivia Watkins, Ethan Adrian Mendes, Justin Svegliato, Luke Bailey, Tiffany Wang, Isaac Ong, Karim Elmaaroufi, Pieter Abbeel, Trevor Darrell, et al. Tensor trust: Interpretable prompt injection attacks from an online game. arXiv preprint arXiv:2311.01011, 2023. · arXiv
  53. Guan Wang, Sijie Cheng, Xianyuan Zhan, Xiangang Li, Sen Song, and Yang Liu. Open- chat: Advancing open-source language models with mixed-quality data. arXiv preprint arXiv:2309.11235, 2023. · arXiv
  54. Hao Wang, Hao Li, Minlie Huang, and Lei Sha. From noise to clarity: Unraveling the adversarial suffix of large language model attacks via translation of text embeddings. arXiv preprint arXiv:2402.16006, 2024. · arXiv
  55. Yihan Wang, Zhouxing Shi, Andrew Bai, and Cho-Jui Hsieh. Defending llms against jailbreak- ing attacks via backtranslation. arXiv preprint arXiv:2402.16459, 2024. · arXiv
  56. Alexander Wei, Nika Haghtalab, and Jacob Steinhardt. Jailbroken: How does llm safety training fail? In Advances in Neural Information Processing Systems (NeurIPS), 2023.
  57. Zeming Wei, Yifei Wang, and Yisen Wang. Jailbreak and guard aligned language models with only few in-context demonstrations. arXiv preprint arXiv:2310.06387, 2023. · arXiv
  58. Yueqi Xie, Jingwei Yi, Jiawei Shao, Justin Curl, Lingjuan Lyu, Qifeng Chen, Xing Xie, and Fangzhao Wu. Defending chatgpt against jailbreak attack via self-reminders. Nature Machine Intelligence, 5(12):1486-1496, 2023.
  59. Zhangchen Xu, Fengqing Jiang, Luyao Niu, Jinyuan Jia, Bill Yuchen Lin, and Radha Pooven- dran. Safedecoding: Defending against jailbreak attacks via safety-aware decoding. arXiv preprint arXiv:2402.08983, 2024. · arXiv
  60. Xianjun Yang, Xiao Wang, Qi Zhang, Linda Petzold, William Yang Wang, Xun Zhao, and Dahua Lin. Shadow alignment: The ease of subverting safely-aligned language models. arXiv preprint arXiv:2310.02949, 2023. · arXiv
  61. Zheng-Xin Yong, Cristina Menghini, and Stephen H Bach. Low-resource languages jailbreak gpt-4. arXiv preprint arXiv:2310.02446, 2023. · arXiv
  62. Youliang Yuan, Wenxiang Jiao, Wenxuan Wang, Jen-tse Huang, Pinjia He, Shuming Shi, and Zhaopeng Tu. Gpt-4 is too smart to be safe: Stealthy chat with llms via cipher. arXiv preprint arXiv:2308.06463, 2023. · arXiv
  63. Zhuowen Yuan, Zidi Xiong, Yi Zeng, Ning Yu, Ruoxi Jia, Dawn Song, and Bo Li. Rigorllm: Resilient guardrails for large language models against undesired content. arXiv preprint arXiv:2403.13031, 2024. · arXiv
  64. Yi Zeng, Hongpeng Lin, Jingwen Zhang, Diyi Yang, Ruoxi Jia, and Weiyan Shi. How johnny can persuade llms to jailbreak them: Rethinking persuasion to challenge ai safety by humanizing llms. arXiv preprint arXiv:2401.06373, 2024. · arXiv
  65. Zhexin Zhang, Junxiao Yang, Pei Ke, and Minlie Huang. Defending large language models against jailbreaking attacks through goal prioritization. arXiv preprint arXiv:2311.09096, 2023. · arXiv
  66. Andy Zhou, Bo Li, and Haohan Wang. Robust prompt optimization for defending language models against jailbreaking attacks. arXiv preprint arXiv:2401.17263, 2024. · arXiv
  67. Yujun Zhou, Yufei Han, Haomin Zhuang, Taicheng Guo, Kehan Guo, Zhenwen Liang, Hongyan Bao, and Xiangliang Zhang. Defending jailbreak prompts via in-context adversarial game. arXiv preprint arXiv:2402.13148, 2024. · arXiv
  68. Sicheng Zhu, Ruiyi Zhang, Bang An, Gang Wu, Joe Barrow, Zichao Wang, Furong Huang, Ani Nenkova, and Tong Sun. Autodan: Automatic and interpretable adversarial attacks on large language models. arXiv preprint arXiv:2310.15140, 2023. · arXiv
  69. Andy Zou, Zifan Wang, J Zico Kolter, and Matt Fredrikson. Universal and transferable adversarial attacks on aligned language models. arXiv preprint arXiv:2307.15043, 2023. · arXiv

Cited by

Sorted by citations

Authors
Zheng, X., Pang, T., Du, C., Liu, Q., Jiang, J., & Lin, M.
ORCID
Venue
Advances in Neural Information Processing Systems 37
Publisher
Neural Information Processing Systems Foundation, Inc. (NeurIPS)
Published
2024
Pages
32856-32887
ISSN
1049-5258 (ISSN-L)
DOI
10.52202/079017-1034
OpenAlex
W4415797913
Cited by
24
References
0
Citations by year
2026: 7 · 2025: 16 · 2024: 1
FWCI
5.80
Type
conference paper
Language
en
Open access
Locations
  • neural information processing systems (publishedVersion)
  • arXiv (Cornell University) (submittedVersion)
  • arXiv (Cornell University)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Adversarial Robustness in Machine Learning
  • Physical Sciences › Computer Science › Artificial Intelligence › Security and Verification in Computing
  • Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
Keywords
large language model security, prompt engineering

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review