Qian Liu

Test-Time Backdoor Attacks on Multimodal Large Language Models

Chao Du, Tianyu Pang, Xianjun Yang, 2024, ArXiv.

Agent Smith: A Single Image Can Jailbreak One Million Multimodal LLM Agents Exponentially Fast

Chao Du, Tianyu Pang, Xiaosen Zheng, 2024, ICML.