ASTPrompter: Preference-Aligned Automated Language Model Red-Teaming to Generate Low-Perplexity Unsafe Prompts

Amelia Hardy, Houjun Liu, Allie Griffith, Bernard Lange, Duncan Eddy, Mykel Kochenderfer


Abstract
Existing LLM red-teaming approaches prioritize high attack success rate, often resulting in high-perplexity prompts. This focus overlooks low-perplexity attacks that are more difficult to filter, more likely to arise during benign usage, and more impactful as negative downstream training examples. In response, we introduce ASTPrompter, a single-step optimization method that uses contrastive preference learning to train an attacker to maintain low perplexity while achieving a high attack success rate (ASR). ASTPrompter achieves an attack success rate 5.1 times higher on Llama-8.1B while using inputs that are 2.1 times more likely to occur according to the frozen LLM. Furthermore, our attack transfers to Mistral-7B, Qwen-7B, and TinyLlama in both black- and white-box settings. Lastly, by tuning a single hyperparameter in our method, we discover successful attack prefixes along an efficient frontier between ASR and perplexity, highlighting perplexity as a previously under-considered factor in red-teaming.
Anthology ID:
2025.findings-emnlp.144
Volume:
Findings of the Association for Computational Linguistics: EMNLP 2025
Month:
November
Year:
2025
Address:
Suzhou, China
Editors:
Christos Christodoulopoulos, Tanmoy Chakraborty, Carolyn Rose, Violet Peng
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
2668–2683
Language:
URL:
https://aclanthology.org/2025.findings-emnlp.144/
DOI:
Bibkey:
Cite (ACL):
Amelia Hardy, Houjun Liu, Allie Griffith, Bernard Lange, Duncan Eddy, and Mykel Kochenderfer. 2025. ASTPrompter: Preference-Aligned Automated Language Model Red-Teaming to Generate Low-Perplexity Unsafe Prompts. In Findings of the Association for Computational Linguistics: EMNLP 2025, pages 2668–2683, Suzhou, China. Association for Computational Linguistics.
Cite (Informal):
ASTPrompter: Preference-Aligned Automated Language Model Red-Teaming to Generate Low-Perplexity Unsafe Prompts (Hardy et al., Findings 2025)
Copy Citation:
PDF:
https://aclanthology.org/2025.findings-emnlp.144.pdf
Checklist:
 2025.findings-emnlp.144.checklist.pdf