Steering Beyond the Support: Adversarial Training on Unsupervised Jailbroken Activation Simulation

Abstract

Jailbreak prompts can trigger harmful completions on aligned LLMs, In accordance, safety steering has been proposed: test-time activation interventions that steer jailbreak activations to trigger refusal while preserving benign utility. However, existing steering methods are fundamentally supervised and tied to a static, limited training set, whereas real jailbreaks evolve and are often out-of-distributed from the training set, leading to failures on unseen attacks. In this paper, we tackle the

Categories

Framework mappings

Suggested from the entry's categories.

@misc{chen2026steering,
  title = {{Steering Beyond the Support: Adversarial Training on Unsupervised Jailbroken Activation Simulation}},
  author = {Luoyu Chen and Weiqi Wang and Zhiyi Tian and Chenhan Zhang and Feng Wu and Jianhuan Huang and Ahmed Asiri and Shui Yu},
  year = {2026},
  month = may,
  eprint = {2605.24535},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2605.24535}
}