Unsafe by Reciprocity: How Generation-Understanding Coupling Undermines Safety in Unified Multimodal Models

TL;DR

This paper uncovers safety vulnerabilities in unified multimodal models caused by their reciprocal understanding and generation capabilities, demonstrating that bidirectional interactions can be exploited to propagate unsafe signals.

Contribution

It introduces RICE, a novel attack framework exploiting cross-functionality reciprocity in UMMs, revealing significant safety risks overlooked in prior research.

Findings

High attack success rates in both understanding-to-generation and generation-to-understanding pathways.

Unsafe signals can propagate across modalities, amplifying safety risks.

Reciprocal coupling in UMMs is a structural vulnerability for safety.

Abstract

Recent advances in Large Language Models (LLMs) and Text-to-Image (T2I) models have led to the emergence of Unified Multimodal Models (UMMs), where multimodal understanding and image generation are tightly integrated within a shared architecture. Prior studies suggest that such reciprocity enhances cross-functionality performance through shared representations and joint optimization. However, the safety implications of this tight coupling remain largely unexplored, as existing safety research predominantly analyzes understanding and generation functionalities in isolation. In this work, we investigate whether cross-functionality reciprocity itself constitutes a structural source of vulnerability in UMMs. We propose RICE: Reciprocal Interaction-based Cross-functionality Exploitation, a novel attack paradigm that explicitly exploits bidirectional interactions between understanding and generation. Using this framework, we systematically evaluate Generation-to-Understanding (G-U) and Understanding-to-Generation (U-G) attack pathways, demonstrating that unsafe intermediate signals can propagate across modalities and amplify safety risks. Extensive experiments show high Attack Success Rates (ASR) in both directions, revealing previously overlooked safety weaknesses inherent to UMMs.