Dissecting Adversarial Robustness of Multimodal LM Agents

Abstract

Demonstrates adversarial attacks on multimodal agents that take actions in digital environments, showing visual perturbations can hijack agent behavior.

Categories

Framework mappings

@inproceedings{wu2025dissecting,
  title = {{Dissecting Adversarial Robustness of Multimodal LM Agents}},
  author = {Chen Henry Wu and Rishi Shah and Jing Yu Koh and Ruslan Salakhutdinov and Daniel Fried and Aditi Raghunathan},
  year = {2025},
  booktitle = {ICLR 2025},
  eprint = {2406.12814},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2406.12814}
}