# From Recognition to Reasoning: Advancing Multimodal Harmful Meme Detection via Chain-of-Thought Alignment

arXiv: https://arxiv.org/abs/2506.18919 · HTML: https://mlsanity.com/paper/2506.18919/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning.
- Builds on 31 of its 59 references that are in the map.
- First posted on arXiv 15 Jun 2025.

## Placed in

- LLM reasoning / [Reasoning benchmarks and evaluation](https://mlsanity.com/lr/reasoning-benchmarks-and-evaluation/)
- LLM safety and security / [Safety alignment, refusal and guardrails](https://mlsanity.com/safety/safety-alignment-refusal-and-guardrails/)

## Builds on (most cited first, 30 of 31)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Scaling Laws for Neural Language Models](https://mlsanity.com/paper/2001.08361/) (2020)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [BERTScore: Evaluating Text Generation with BERT](https://mlsanity.com/paper/1904.09675/) (2019)
- [GPT-4o System Card](https://mlsanity.com/paper/2410.21276/) (2024)
- [InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models](https://mlsanity.com/paper/2504.10479/) (2025)
- [Understanding deep learning requires rethinking generalization](https://mlsanity.com/paper/1611.03530/) (2016)
- [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://mlsanity.com/paper/1908.03557/) (2019)
- [MiniCPM-V: A GPT-4V Level MLLM on Your Phone](https://mlsanity.com/paper/2408.01800/) (2024)
- [Deep Learning Scaling is Predictable, Empirically](https://mlsanity.com/paper/1712.00409/) (2017)
- [The Hateful Memes Challenge: Detecting Hate Speech in Multimodal Memes](https://mlsanity.com/paper/2005.04790/) (2020)
- [MOMENTA: A Multimodal Framework for Detecting Harmful Memes and Their Targets](https://mlsanity.com/paper/2109.05184/) (2021)
- [Detecting Harmful Memes and Their Targets](https://mlsanity.com/paper/2110.00413/) (2021)
- [Prompting for Multimodal Hateful Meme Classification](https://mlsanity.com/paper/2302.04156/) (2023)
- [Hate-CLIPper: Multimodal Hateful Meme Classification based on Cross-modal Interaction of CLIP Features](https://mlsanity.com/paper/2210.05916/) (2022)
- [Pro-Cap: Leveraging a Frozen Vision-Language Model for Hateful Meme Detection](https://mlsanity.com/paper/2308.08088/) (2023)
- [MemeCLIP: Leveraging CLIP Representations for Multimodal Meme Classification](https://mlsanity.com/paper/2409.14703/) (2024)
- [Hate Speech in Pixels: Detection of Offensive Memes towards Automatic Moderation](https://mlsanity.com/paper/1910.02334/) (2019)
- [On Explaining Multimodal Hateful Meme Detection Models](https://mlsanity.com/paper/2204.01734/) (2022)
- [Towards Low-Resource Harmful Meme Detection with LMM Agents](https://mlsanity.com/paper/2411.05383/) (2024)
- [Improving Hateful Meme Detection through Retrieval-Guided Contrastive Learning](https://mlsanity.com/paper/2311.08110/) (2023)
- [DISARM: Detecting the Victims Targeted by Harmful Memes](https://mlsanity.com/paper/2205.05738/) (2022)
- [Mapping Memes to Words for Multimodal Hateful Meme Classification](https://mlsanity.com/paper/2310.08368/) (2023)
- [Towards Explainable Harmful Meme Detection through Multimodal Debate between Large Language Models](https://mlsanity.com/paper/2401.13298/) (2024)
- [Cross-Dataset Adaptation for Visual Question Answering](https://mlsanity.com/paper/1806.03726/) (2018)
- [Towards Comprehensive Detection of Chinese Harmful Memes](https://mlsanity.com/paper/2410.02378/) (2024)
- [Robust Adaptation of Large Multimodal Models for Retrieval Augmented Hateful Meme Detection](https://mlsanity.com/paper/2502.13061/) (2025)
- [Read as You See: Guiding Unimodal LLMs for Low-Resource Explainable Harmful Meme Detection](https://mlsanity.com/paper/2506.08477/) (2025)

## Built on by (most cited first, 4 of 4)

- [Labels or Input? Rethinking Augmentation in Multimodal Hate Detection](https://mlsanity.com/paper/2508.11808/) (2025)
- [Is GPT-4o mini Blinded by its Own Safety Filters? Exposing the Multimodal-to-Unimodal Bottleneck in Hate Speech Detection](https://mlsanity.com/paper/2509.13608/) (2025)
- [Distinguishing Right from Wrong in Debates: Attribution Analysis of Chinese Harmful Memes](https://mlsanity.com/paper/2605.24344/) (2026)
- [HarmTrace: Anchor-Calibrated Decoupled Optimization for Fine-Grained Target Identification in Harmful Memes](https://mlsanity.com/paper/2608.16622/) (2026)

## Abstract (arXiv)

As a multimodal communication medium that integrates images and text, memes often convey implicit harmful content through metaphors, satire, and humor, making harmful meme detection a complex and challenging task. Although recent studies have achieved considerable progress in detection accuracy and model interpretability, large-scale, high-quality datasets for harmful memes remain scarce. Moreover, existing methods still exhibit notable limitations in identifying implicit risks and understanding fine-grained semantics. To address these challenges, we construct MemeMind, a large-scale dataset for harmful meme detection. MemeMind comprises a broad collection of publicly available memes and adopts a rigorous and comprehensive taxonomy of harmful content developed in accordance with widely recognized international standards and contemporary Internet contexts. In addition, the dataset provides detailed structured Chain-of-Thought (CoT) reasoning annotations to support fine-grained analysis of harmfulness, implicit intentions, and underlying semantics in memes. Building upon MemeMind, we further propose MemeGuard, a reasoning-oriented multimodal framework for harmful meme detection. MemeGuard employs a three-stage training strategy to progressively enhance the model's visual understanding, multimodal reasoning, and harmful content discrimination capabilities, thereby improving both detection accuracy and the interpretability of model decisions. Extensive experimental results demonstrate that MemeGuard outperforms existing state-of-the-art methods on the MemeMind dataset, providing a solid foundation for future research on harmful meme detection and multimodal content safety.