When World Models Dream Wrong: Physical-Conditioned Adversarial Attacks against World Models

This research introduces PhysCond-WMA, a novel white-box adversarial attack targeting the physical-conditioned channels of generative world models, revealing...

Level: advanced

By Zhixiang Guo, Siyuan Liang, Andras Balogh, Noah Lunberry, Rong-Cheng Tu, Mark Jelasity, Dacheng Tao

Category: research