Jailbreaking Black Box Large Language Models in Twenty Queries
Open accessThe DOI matches its Crossref record; 36 of the 36 references it cites were found in a registry or an index.
Abstract
There is growing interest in ensuring that large language models (LLMs) align with human values. However, the alignment of such models is vulnerable to adversarial jailbreaks, which coax LLMs into overriding their safety guardrails. The identification of these vulnerabilities is therefore instrumental in understanding inherent weaknesses and preventing future misuse. To this end, we propose Prompt Automatic Iterative Refinement (PAIR), an algorithm that generates semantic jailbreaks with only black-box access to an LLM. PAIR—which is inspired by social engineering attacks—uses an attacker LLM to automatically generate jailbreaks for a separate targeted LLM without human intervention. In this way, the attacker LLM iteratively queries the target LLM to update and refine a candidate jailbreak. Empirically, PAIR often requires fewer than twenty queries to produce a jailbreak, which is orders of magnitude more efficient than existing algorithms. PAIR also achieves competitive jailbreaking success rates and transferability on open and closed-source LLMs, including GPT-3.5/4, Vicuna, and Gemini.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Language Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372 · cited 1×
“The strong performance of LLMs is largely due to the fact that they are trained on vast text corpora—often scraped from the Internet—which in turn facilitates the generation of realistic text that pertains to a diverse set of topics [5, 33].”
Explaining and Harnessing Adversarial Examples
Goodfellow, I. J., Shlens, J., & Szegedy, C. (2014) · arXiv · doi:10.48550/arxiv.1412.6572 · Cited by 10,381
Intriguing properties of neural networks
Szegedy, C., Zaremba, W., Sutskever, I., Bruna, J., Erhan, D., Goodfellow, I., & Fergus, R. (2013) · arXiv · doi:10.48550/arxiv.1312.6199 · Cited by 8,143
Towards Deep Learning Models Resistant to Adversarial Attacks
Madry, A., Makelov, A., Schmidt, L., Tsipras, D., & Vladu, A. (2017) · arXiv · doi:10.48550/arxiv.1706.06083 · Cited by 6,602
Chain-Of-Thought Prompting Elicits Reasoning in Large Language Models
Wei, J., Wang, X., Schuurmans, D., Bosma, M., Ichter, B., Xia, F., Chi, E., Le, Q. V., & Zhou, D. (2022) · Advances in Neural Information Processing Systems 35 · doi:10.52202/068431-1800 · Cited by 4,505
Large language models in medicine
Thirunavukarasu, A. J., Ting, D. S. J., Elangovan, K., Gutierrez, L., Tan, T. F., & Ting, D. S. W. (2023) · Nature Medicine · doi:10.1038/s41591-023-02448-8 · Cited by 3,944
Llama 2: Open Foundation and Fine-Tuned Chat Models
Touvron, H., Martin, L., Stone, K., Albert, P., Almahairi, A., Babaei, Y., Bashlykov, N., Batra, S., Bhargava, P., Bhosale, S., et al. (2023) · arXiv · doi:10.48550/arxiv.2307.09288 · Cited by 2,604
Making Pre-trained Language Models Better Few-shot Learners
Gao, T., Fisch, A., & Chen, D. (2021) · Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) · doi:10.18653/v1/2021.acl-long.295 · Cited by 1,380
Delving into Transferable Adversarial Examples and Black-box Attacks
Liu, Y., Chen, X., Liu, C., & Song, D. (2016) · arXiv · doi:10.48550/arxiv.1611.02770 · Cited by 1,310
AutoPrompt: Eliciting Knowledge from Language Models with Automatically Generated Prompts
Shin, T., Razeghi, Y., Logan IV, R. L., Wallace, E., & Singh, S. (2020) · Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2020.emnlp-main.346 · Cited by 1,279
Research works printed but not linked: 29
- Wu, 2023, Bloomberggpt: A large language model for finance, arXiv preprint
- Korbak, 2023, Pretraining language models with human preferences, 17506
- Ouyang, 2022, Training language models to follow instructions with human feedback, 2022, 13
- Carlini, 2023, Are aligned neural networks adversarially aligned?
- Maus, 2023, Black box adversarial prompting for foundation models
- Jones, 2023, Automatically auditing large language models via discrete optimization
- Zou, 2023, Universal and transferable adversarial attacks on aligned language models
- Inan, 2023, Llama guard: Llm-based input-output safe-guard for human-ai conversations
- Bai, 2022, Constitutional ai: Harmlessness from ai feedback
- 10.18653/v1/2021.emnlp-main.696 · DOI · Paper page
- Bartolo, 2021, Models in the loop: Aiding crowdworkers with generative annotation assistants, arXiv preprint
- Huang, 2023, Catastrophic jailbreak of open-source llms via exploiting generation
- Mazeika, 2023, Tdc 2023 (llm edition): The trojan detection challenge
- Jiang, 2024, Mixtral of experts
- Zheng, 2023, Judging llm-as-a-judge with mt-bench and chatbot arena
- 2023, Gpt-4 technical report
- 2023, Gemini: A family of highly capable multi-modal models
- Jain, 2023, Baseline defenses for adversarial attacks against aligned language models, arXiv preprint
- Alon, 2023, Detecting language model attacks with perplexity, arXiv preprint
- 10.1007/978-3-319-94042-7_11 · DOI · Paper page
- Wang, 2019, Improving adversarial robustness requires revisiting misclassified examples, International conference on learning representations
- 10.1109/SP.2019.00044 · DOI · Paper page
- Cohen, 2019, Certified adversarial robustness via randomized smoothing, 1310
- Salman, 2019, Provably robust deep learning via adversarially trained smoothed classifiers, Advances in Neural Information Processing Systems, 32
- 10.14722/ndss.2024.24188 · DOI · Paper page
- 10.1109/iccv48922.2021.00760 · DOI
- 10.1145/3128572.3140448 · DOI · Paper page
- Ji, 2024, Defending large language models against jailbreak attacks via semantic smoothing, arXiv preprint
- Zhou, 2024, Robust prompt optimization for defending language models against jailbreaking attacks, arXiv preprint
Cited by
Security and Privacy Challenges of Large Language Models: A Survey
Das, B. C., Amini, M. H., & Wu, Y. (2025) · ACM Computing Surveys · doi:10.1145/3712001 · Cited by 445
“It can automatically generate jailbreaking prompts for a distinct targeted LLM, eliminating the need for human intervention [79].”
Jailbreaking Leading Safety-Aligned LLMs with Simple Adaptive Attacks
Andriushchenko, M., Croce, F., & Flammarion, N. (2024) · arXiv · doi:10.48550/arxiv.2404.02151 · Cited by 396
AdvPrompter: Fast Adaptive Adversarial Prompting for LLMs
Paulus, A., Zharmagambetov, A., Guo, C., Amos, B., & Tian, Y. (2024) · arXiv · doi:10.48550/arxiv.2404.16873 · Cited by 184
Zeng, Y., Lin, H., Zhang, J., Yang, D., Jia, R., & Shi, W. (2024) · arXiv · doi:10.48550/arxiv.2401.06373 · Cited by 174
Hughes, J., Price, S., Lynch, A., Schaeffer, R., Barez, F., Koyejo, S., Sleight, H., Jones, E., Perez, E., & Sharma, M. (2024) · arXiv · doi:10.48550/arxiv.2412.03556 · Cited by 129
Metadata
- Authors
- Chao, P., Robey, A., Dobriban, E., Hassani, H., Pappas, G. J., & Wong, E.
- Venue
- 2025 IEEE Conference on Secure and Trustworthy Machine Learning (SaTML)
- Publisher
- IEEE
- Published
- 2025-04-09
- Pages
- 23-42
- OpenAlex
- W4410609100
- Cited by
- 193
- References
- 36
- FWCI
- 175.49
- Type
- conference paper
- Language
- en
- Open access
- green
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
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Jailbreaking Black Box Large Language Models in Twenty Queries. https://app.cochat.ai/papers/10.1109/satml64287.2025.00010
- The paper
- Chao, P., Robey, A., Dobriban, E., Hassani, H., Pappas, G. J., & Wong, E. (2025). Jailbreaking Black Box Large Language Models in Twenty Queries. 2025 IEEE Conference on Secure and Trustworthy Machine Learning (SaTML). https://doi.org/10.1109/satml64287.2025.00010
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.