Ellipsoid Control: A White-list Jailbreak Defense via Benign Latent Modeling

Abstract

Representation engineering (RepE) defenses have shown strong robustness against jailbreak attacks on large language models (LLMs). However, these methods fundamentally rely on black-list supervision: they learn jailbreak-to-refusal activation transformations from harmful or jailbreak data that are inherently incomplete and continuously evolving. Hence, the performance of RepE-based defenses becomes tightly coupled to the quality and coverage of collected harmful samples, leaving models vulnerabl

Categories

Framework mappings

Suggested from the entry's categories.

@misc{chen2026ellipsoid,
  title = {{Ellipsoid Control: A White-list Jailbreak Defense via Benign Latent Modeling}},
  author = {Luoyu Chen and Weiqi Wang and Zhiyi Tian and Feng Wu and Ahmed Asiri and Shui Yu},
  year = {2026},
  month = may,
  eprint = {2605.24552},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2605.24552}
}