Is AI Ready for Multimodal Hate Speech Detection? A Comprehensive Dataset and Benchmark Evaluation

TL;DR

This paper introduces a new multimodal hate speech dataset and evaluates large language models, revealing their struggles with contextual reasoning in meme-based hate speech detection.

Contribution

It presents M^3, a fine-grained, multi-platform multimodal hate speech dataset with hierarchical labels and rationales, and benchmarks state-of-the-art models highlighting their limitations.

Findings

Models struggle with context utilization in memes.

Surrounding discourse often degrades detection performance.

Highlighting the need for context-aware multimodal architectures.

Abstract

Hate speech online targets individuals or groups based on identity attributes and spreads rapidly, posing serious social risks. Memes, which combine images and text, have emerged as a nuanced vehicle for disseminating hate speech, often relying on cultural knowledge for interpretation. However, existing multimodal hate speech datasets suffer from coarse-grained labeling and a lack of integration with surrounding discourse, leading to imprecise and incomplete assessments. To bridge this gap, we propose an agentic annotation framework that coordinates seven specialized agents to generate hierarchical labels and rationales. Based on this framework, we construct M^3 (Multi-platform, Multi-lingual, and Multimodal Meme), a dataset of 2,455 memes collected from X, 4chan, and Weibo, featuring fine-grained hate labels and human-verified rationales. Benchmarking state-of-the-art Multimodal Large Language Models reveals that these models struggle to effectively utilize surrounding post context, which often fails to improve or even degrades detection performance. Our finding highlights the challenges these models face in reasoning over memes embedded in real-world discourse and underscores the need for a context-aware multimodal architecture. Our dataset and code are available at https://github.com/mira-ai-lab/M3.