Breaking Cross-modal Alignment in Embodied Intelligence: A Multimodal Adversarial Attack Framework for Vision-Language-Action Models.

@inproceedings{zhao-2026-breaking, author = {Zhihui Zhao and Xiaorong Dong and Yaowen Zheng and Xiaohui Chen and Yimo Ren and Hangbei Cheng and Yongle Chen and Limin Sun}, editor = {Hakim Hacid and Yoelle Maarek and Francesco Bonchi and Ido Guy and Emine Yilmaz}, title = {{Breaking Cross-modal Alignment in Embodied Intelligence: A Multimodal Adversarial Attack Framework for Vision-Language-Action Models}}, booktitle = {{Proceedings of the ACM Web Conference 2026, WWW 2026, Dubai, United Arab Emirates, originally scheduled for April 13-17, 2026, rescheduled for June 29 - July 3, 2026}}, pages = {2824--2834}, publisher = {ACM}, year = {2026}, url = {https://doi.org/10.1145/3774904.3792315}, doi = {https://doi.org/10.1145/3774904.3792315} }