Jailbreak susceptibility prediction and mitigation via the behavioral geometry of models

Abstract

Evaluating and mitigating a generative system's susceptibility to jailbreak attacks is critical to its safe deployment. Given the number of deployable systems, full per-configuration evaluation and optimization is impractical. In this paper, we formalize the behavioral geometry of a population of models that, by leveraging previously evaluated and defended models, supports both efficient susceptibility prediction and effective defense transfer across a population. We apply the framework to 79 mo

Categories

Framework mappings

Suggested from the entry's categories.

@misc{helm2026jailbreak,
  title = {{Jailbreak susceptibility prediction and mitigation via the behavioral geometry of models}},
  author = {Hayden Helm and Xiaodong Liu and Weiwei Yang},
  year = {2026},
  month = may,
  eprint = {2605.26409},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2605.26409}
}