作者位次 #3/7ICLR 2026Poster
Toward Universal and Transferable Jailbreak Attacks on Vision-Language Models
Kaiyuan Cui · Yige Li · Yutao Wu · Xingjun Ma · Sarah Erfani · Christopher Leckie · Hanxun Huang
首次获取该候选人的完整档案及 StartLux 评估共 20 积分；每位用户单独解锁，之后查看免费。主动刷新 20 积分。
实时查找电子邮箱，费用不超过20积分；未找到邮箱不扣费。
Kaiyuan Cui · Yige Li · Yutao Wu · Xingjun Ma · Sarah Erfani · Christopher Leckie · Hanxun Huang
Yunhao Feng · Yige Li · Yutao Wu · Yingshui Tan · Yanming Guo · Yifan Ding · Kun Zhai · Xingjun Ma · Yu-Gang Jiang
Large language model (LLM) agents execute tasks through multi-step workflows that combine planning, memory, and tool use. While this design enables autonomy, it also expands the attack surface for backdoor threats. Backdoor triggers injected into specific stages of an agent workflow can persist t...