CrossGuard: Safeguarding MLLMs against Joint-Modal Implicit Malicious Attacks

TL;DR

This paper introduces CrossGuard, a novel defense system for multimodal large language models that effectively detects and mitigates both explicit and implicit joint-modal malicious attacks, enhancing robustness in real-world scenarios.

Contribution

The paper presents ImpForge, an automated pipeline for generating implicit attack data, and CrossGuard, a new intent-aware safeguard that outperforms existing defenses against multimodal threats.

Findings

CrossGuard significantly outperforms existing defenses in various benchmarks.

ImpForge generates diverse implicit attack samples across 14 domains.

CrossGuard maintains high utility while providing robust security.

Abstract

Multimodal Large Language Models (MLLMs) achieve strong reasoning and perception capabilities but are increasingly vulnerable to jailbreak attacks. While existing work focuses on explicit attacks, where malicious content resides in a single modality, recent studies reveal implicit attacks, in which benign text and image inputs jointly express unsafe intent. Such joint-modal threats are difficult to detect and remain underexplored, largely due to the scarcity of high-quality implicit data. We propose ImpForge, an automated red-teaming pipeline that leverages reinforcement learning with tailored reward modules to generate diverse implicit samples across 14 domains. Building on this dataset, we further develop CrossGuard, an intent-aware safeguard providing robust and comprehensive defense against both explicit and implicit threats. Extensive experiments across safe and unsafe benchmarks, implicit and explicit attacks, and multiple out-of-domain settings demonstrate that CrossGuard significantly outperforms existing defenses, including advanced MLLMs and guardrails, achieving stronger security while maintaining high utility. This offers a balanced and practical solution for enhancing MLLM robustness against real-world multimodal threats. Our code is released: https://github.com/ZhangXu0963/CrossGuard.