As large language models (LLMs) are becoming more capable and widespread, the study of their failure cases is becoming increasingly important. Recent advances in standardizing, measuring, and scaling test-time compute suggest new methodologies for optimizing models to achieve high performance on hard tasks. In this paper, we apply these advances to the task of model jailbreaking: eliciting harmful responses from aligned LLMs. We develop an adversarial reasoning approach to automatic jailbreaking that leverages a loss signal to guide the test-time compute, achieving SOTA attack success rates against many aligned LLMs, even those that aim to trade inference-time compute for adversarial robustness. Our approach introduces a new paradigm in understanding LLM vulnerabilities, laying the foundation for the development of more robust and trustworthy AI systems.
@article{arxiv.2502.01633,
title = {Adversarial Reasoning at Jailbreaking Time},
author = {Mahdi Sabbaghi and Paul Kassianik and George Pappas and Yaron Singer and Amin Karbasi and Hamed Hassani},
journal= {arXiv preprint arXiv:2502.01633},
year = {2025}
}
Comments
Accepted to the 42nd International Conference on Machine Learning (ICML 2025)