ShieldVLM: Safeguarding the Multimodal Implicit Toxicity via Deliberative Reasoning with LVLMs

TL;DR

This paper introduces ShieldVLM, a model that uses deliberative reasoning to detect implicit toxicity in multimodal text-image content, addressing a previously underexplored challenge in toxicity moderation.

Contribution

It presents a comprehensive taxonomy and dataset for multimodal implicit toxicity, and develops ShieldVLM, the first model to effectively identify implicit toxicity through cross-modal reasoning.

Findings

ShieldVLM outperforms existing baselines in toxicity detection.

The dataset covers 7 risk categories and 5 cross-modal modes.

The approach enhances detection of both implicit and explicit toxicity.

Abstract

Toxicity detection in multimodal text-image content faces growing challenges, especially with multimodal implicit toxicity, where each modality appears benign on its own but conveys hazard when combined. Multimodal implicit toxicity appears not only as formal statements in social platforms but also prompts that can lead to toxic dialogs from Large Vision-Language Models (LVLMs). Despite the success in unimodal text or image moderation, toxicity detection for multimodal content, particularly the multimodal implicit toxicity, remains underexplored. To fill this gap, we comprehensively build a taxonomy for multimodal implicit toxicity (MMIT) and introduce an MMIT-dataset, comprising 2,100 multimodal statements and prompts across 7 risk categories (31 sub-categories) and 5 typical cross-modal correlation modes. To advance the detection of multimodal implicit toxicity, we build ShieldVLM, a model which identifies implicit toxicity in multimodal statements, prompts and dialogs via deliberative cross-modal reasoning. Experiments show that ShieldVLM outperforms existing strong baselines in detecting both implicit and explicit toxicity. The model and dataset will be publicly available to support future researches. Warning: This paper contains potentially sensitive contents.