paperSeptember 2026Unreviewed

AgentHijack: Visual Patch Attacks on Multimodal Computer-Use Agents

Zhihao Liu, Hongyu Sun, Zhiyuan Fu, Xiaonan Duan, Jice Wang, Shangru Zhao, Weizhi Meng, Wuxin Yang, Yangfan Zhou, Yuqing Zhang

View resource PDF

Abstract

This paper presents an end-to-end evaluation framework for image-triggered command injection against computer-use agents (CUAs). The goal is to test whether a local visual patch can induce verifiable environmental consequences along the full chain of screenshot input, VLM generation, action parsing, and environment execution. We train and deploy patches on author-controlled GitHub Pages pages and a locally deployed CSDN clone, and evaluate them in real environments across five open-source or pub

Cite

@misc{liu2026agenthijack,
  title = {{AgentHijack: Visual Patch Attacks on Multimodal Computer-Use Agents}},
  author = {Zhihao Liu and Hongyu Sun and Zhiyuan Fu and Xiaonan Duan and Jice Wang and Shangru Zhao and Weizhi Meng and Wuxin Yang and Yangfan Zhou and Yuqing Zhang},
  year = {2026},
  month = sep,
  eprint = {2609.09212},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.09212}
}