# HarmVideoBench: Benchmarking Harmful Video Understanding in Large Multimodal Models

arXiv: https://arxiv.org/abs/2606.27187 · HTML: https://mlsanity.com/paper/2606.27187/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in MLLM jailbreaks and adversarial robustness, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 19 of its 32 references that are in the map.
- First posted on arXiv 25 Jun 2026.

## Placed in

- Multimodal LLMs / vision-language models / [MLLM jailbreaks and adversarial robustness](https://mlsanity.com/mllm/mllm-jailbreaks-and-adversarial-robustness/)

## Builds on (most cited first, 19 of 19)

- [Qwen3-VL Technical Report](https://mlsanity.com/paper/2511.21631/) (2025)
- [Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysis](https://mlsanity.com/paper/2405.21075/) (2024)
- [MVBench: A Comprehensive Multi-modal Video Understanding Benchmark](https://mlsanity.com/paper/2311.17005/) (2023)
- [Phi-4 Technical Report](https://mlsanity.com/paper/2412.08905/) (2024)
- [LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understanding](https://mlsanity.com/paper/2407.15754/) (2024)
- [ActivityNet-QA: A Dataset for Understanding Complex Web Videos via Question Answering](https://mlsanity.com/paper/1906.02467/) (2019)
- [Kimi K2.5: Visual Agentic Intelligence](https://mlsanity.com/paper/2602.02276/) (2026)
- [HateXplain: A Benchmark Dataset for Explainable Hate Speech Detection](https://mlsanity.com/paper/2012.10289/) (2020)
- [MM-SafetyBench: A Benchmark for Safety Evaluation of Multimodal Large Language Models](https://mlsanity.com/paper/2311.17600/) (2023)
- [SPA-VL: A Comprehensive Safety Preference Alignment Dataset for Vision Language Model](https://mlsanity.com/paper/2406.12030/) (2024)
- [Reasoning Over Paragraph Effects in Situations](https://mlsanity.com/paper/1908.05852/) (2019)
- [WIQA: A dataset for "What if..." reasoning over procedural text](https://mlsanity.com/paper/1909.04739/) (2019)
- [HateMM: A Multi-Modal Dataset for Hate Video Classification](https://mlsanity.com/paper/2305.03915/) (2023)
- [Video-SafetyBench: A Benchmark for Safety Evaluation of Video LVLMs](https://mlsanity.com/paper/2505.11842/) (2025)
- [MultiHateClip: A Multilingual Benchmark Dataset for Hateful Video Detection on YouTube and Bilibili](https://mlsanity.com/paper/2408.03468/) (2024)
- [Can't See the Forest for the Trees: Benchmarking Multimodal Safety Awareness for Multimodal LLMs](https://mlsanity.com/paper/2502.11184/) (2025)
- [Beneath the Surface: Unveiling Harmful Memes with Multimodal Reasoning Distilled from Large Language Models](https://mlsanity.com/paper/2312.05434/) (2023)
- [Lance: Unified Multimodal Modeling by Multi-Task Synergy](https://mlsanity.com/paper/2605.18678/) (2026)
- [SafeVid: Toward Safety Aligned Video Large Multimodal Models](https://mlsanity.com/paper/2505.11926/) (2025)

## Abstract (arXiv)

Large vision-language models (LVLMs) have recently shown immense potential in automated content moderation, sparking growing interest in developing harmful-video benchmarks. However, we identify two primary limitations in existing works: 1) The multi-layered characteristics of harmful videos are overlooked. Existing benchmarks predominantly formulate evaluation as a binary classification task, failing to capture implicit or deep contextual harms. 2) Explanatory rationales are completely absent. Current frameworks measure exclusively whether a model flags a video correctly rather than explaining why, turning evaluation into a black box where models can succeed through superficial shortcuts. To address these problems, we present HarmVideoBench, a multi-layered diagnostic benchmark comprising 1,379 videos paired with 4,137 multiple-choice questions. HarmVideoBench benchmarks three hierarchical dimensions: Observable Evidence, Clip-Internal Meaning, and Beyond-Clip Reasoning, aiming to evaluate models' deep understanding beyond surface cues with carefully balanced and curated samples. We evaluate 19 leading models on HarmVideoBench to assess their multidimensional understanding of harmful videos. Moreover, we introduce BCR, a benchmark-aligned method that predicts reasoning boundaries and dynamically retrieves context only when needed. Experimental results show that BCR substantially improves the base model's performance in harmful video understanding, raising the macro average from 61.7 percent to a state-of-the-art 84.4 percent.