作者位次 #6/9ACL 2026Findings· Findings of ACL
EVA : Evolving Semantic Adversaries for Red-Teaming GUI Agents Against Environmental Injection Attacks
Yijie Lu · Manman Zhao · Tianjie Ju · Zihe Yan · Xinbei Ma · Yuan Guo · Daizong Ding · Gongshen Liu · Zhuosheng Zhang
摘要
Autonomous GUI agents are inherently vulnerable to Environmental Injection Attacks (EIAs). However, existing red-teaming methods face a trade-off between requiring target-specific knowledge and incurring prohibitive computational costs. More fundamentally, a key question remains: \textit{what fac...