When One Modality Rules Them All: Backdoor Modality Collapse in Multimodal Diffusion Models
This research identifies a critical blind spot in understanding backdoor vulnerabilities in multimodal diffusion models, which is important for developers and users of these models to build more robust systems.
This paper investigates backdoor attacks in multimodal diffusion models, specifically focusing on the phenomenon of "Backdoor Modality Collapse" where attacks predominantly rely on a subset of modalities. The authors found that attacks often collapse into subset-modality dominance and that cross-modal interaction is negligible or even negative, challenging the assumption of synergistic vulnerability.
While diffusion models have revolutionized visual content generation, their rapid adoption has underscored the critical need to investigate vulnerabilities, e.g., to backdoor attacks. In multimodal diffusion models, it is natural to expect that attacking multiple modalities simultaneously (e.g., text and image) would yield complementary effects and strengthen the overall backdoor. In this paper, we challenge this assumption by investigating the phenomenon of Backdoor Modality Collapse, a scenario where the backdoor mechanism degenerates to rely predominantly on a subset of modalities, rendering others redundant. To rigorously quantify this behavior, we introduce two novel metrics: Trigger Modality Attribution (TMA) and Cross-Trigger Interaction (CTI). Through extensive experiments across diverse training configurations in multimodal conditional diffusion, we consistently observe a ``winner-takes-all'' dynamic in backdoor behavior. Our results reveal that (1) attacks often collapse into subset-modality dominance, and (2) cross-modal interaction is negligible or even negative, contradicting the intuition of synergistic vulnerability. These findings highlight a critical blind spot in current assessments, suggesting that high attack success rates often mask a fundamental reliance on a subset of modalities. This establishes a principled foundation for mechanistic analysis and future defense development.