Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces ContrAR, a benchmark dataset of 312 real-world AR videos with contradictory virtual content, to evaluate the robustness of vision-language models (VLMs) in AR environments. Eleven VLMs, including commercial and open-source models, were tested on their ability to detect and reason about adversarial content manipulations. Results show that while VLMs demonstrate some understanding of contradictory content, significant improvements are needed, particularly in balancing detection accuracy and latency.
AR's increasing integration into daily life is creating a ripe attack surface: current vision-language models struggle to detect and reason about contradictory virtual content, leaving users vulnerable to manipulation.
Augmented reality (AR) has rapidly expanded over the past decade. As AR becomes increasingly integrated into daily life, its security and reliability emerge as critical challenges. Among various threats, contradictory virtual content attacks, where malicious or inconsistent virtual elements are introduced into the user's view, pose a unique risk by misleading users, creating semantic confusion, or delivering harmful information. In this work, we systematically model such attacks and present ContrAR, a novel benchmark for evaluating the robustness of vision-language models (VLMs) against virtual content manipulation and contradiction in AR. ContrAR contains 312 real-world AR videos validated by 10 human participants. We further benchmark 11 VLMs, including both commercial and open-source models. Experimental results reveal that while current VLMs exhibit reasonable understanding of contradictory virtual content, room still remains for improvement in detecting and reasoning about adversarial content manipulations in AR environments. Moreover, balancing detection accuracy and latency remains challenging.