TempJail: Temporal Jailbreak Attack against Large Vision-Language Models via Subtitle Scheduling

Abstract

Large vision-language models (LVLMs) have achieved remarkable progress in video understanding and reasoning. Despite extensive studies on text- and image-based jailbreaks, video jailbreaks against LVLMs remain largely unexplored. Existing video jailbreak methods mainly manipulate textual content embedded in videos, while overlooking how such information is organized over time. Our analysis reveals that jailbreak effectiveness depends not only on the semantics of textual information but also on i

Categories

Framework mappings

Suggested from the entry's categories.

@misc{zhou2026tempjail,
  title = {{TempJail: Temporal Jailbreak Attack against Large Vision-Language Models via Subtitle Scheduling}},
  author = {Ling Zhou and Yihao Huang and Jinglin Sun and Zhiwen Tian and Yi Zeng and Qi-He Liu and Shi-Jie Zhou},
  year = {2026},
  month = aug,
  eprint = {2608.19737},
  archivePrefix = {arXiv},
  url = {https://www.semanticscholar.org/paper/9256fdf3d5d0f55a454ebf4d96a30155cdf62b38}
}