Training-Free and Interpretable Hateful Video Detection via Multi-stage Adversarial Reasoning

2026-01-21

AI Summary

This work addresses the limitations of existing hate video detection methods, which suffer from scarce training data, poor interpretability, and unreliable performance when directly employing large vision-language models (LVLMs). To overcome these challenges, the authors propose MARS, a novel framework that introduces, for the first time, a multi-stage adversarial reasoning mechanism to achieve high accuracy and strong interpretability without any model training. MARS neutrally describes video content and concurrently generates evidence both supporting and opposing a hate label, then fuses these perspectives to produce an interpretable final judgment. Evaluated on two real-world datasets, MARS substantially outperforms current training-free approaches—by up to 10% in accuracy—and even surpasses the best trained method on one dataset, while providing human-understandable rationales that enhance transparency and compliance in content moderation.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Hateful videos pose serious risks by amplifying discrimination, inciting violence, and undermining online safety. Existing training-based hateful video detection methods are constrained by limited training data and lack of interpretability, while directly prompting large vision-language models often struggle to deliver reliable hate detection. To address these challenges, this paper introduces MARS, a training-free Multi-stage Adversarial ReaSoning framework that enables reliable and interpretable hateful content detection. MARS begins with the objective description of video content, establishing a neutral foundation for subsequent analysis. Building on this, it develops evidence-based reasoning that supports potential hateful interpretations, while in parallel incorporating counter-evidence reasoning to capture plausible non-hateful perspectives. Finally, these perspectives are synthesized into a conclusive and explainable decision. Extensive evaluation on two real-world datasets shows that MARS achieves up to 10% improvement under certain backbones and settings compared to other training-free approaches and outperforms state-of-the-art training-based methods on one dataset. In addition, MARS produces human-understandable justifications, thereby supporting compliance oversight and enhancing the transparency of content moderation workflows. The code is available at https://github.com/Multimodal-Intelligence-Lab-MIL/MARS.

Problem

• hateful video detection
• training-free
• interpretability
• large vision-language models
• content moderation

Innovation

• training-free
• interpretable AI
• adversarial reasoning
• hateful video detection
• vision-language models