Extended to Reality: Prompt Injection in 3D Environments

Zhuoheng Li, Ying Chen

Multimodal large language models (MLLMs) have advanced the capabilities to interpret and act on visual input in 3D environments, empowering diverse applications such as robotics an…