The authors built a high-fidelity interactive mobile evaluation framework that simulates dynamic, cross-application smartphone environments.
Mobile agents are broadly vulnerable to cross-modal injection attacks that hijack reasoning
A new benchmark, MobileInject, shows that attackers can hide malicious objectives across visual and textual mobile content to redirect agent decisions and trigger unsafe actions.
Chinese Tech
Songping Wang · Peiyang Li · Yi Huang · Fengting Li · Xuhao Ran · Yueming Lyu · +2 more
Nanjing University · Ant Group · Peking University
Research Digest··2 min read
The authors introduce MobileInject, the first benchmark for evaluating multimodal large language model based mobile agents against indirect prompt injection attacks.
§