Detecting and Mitigating Hateful Content in Multimodal Memes with Vision-Language Models

• URL: http://arxiv.org/abs/2505.00150v1
• Date: Wed, 30 Apr 2025 19:48:12 GMT
• Title: Detecting and Mitigating Hateful Content in Multimodal Memes with Vision-Language Models
• Authors: Minh-Hao Van, Xintao Wu,
• Abstract summary: Multimodal memes are sometimes misused to disseminate hate speech against individuals or groups.<n>We propose a definition-guided prompting technique for detecting hateful memes, and a unified framework for mitigating hateful content in memes, named UnHateMeme.<n>Our framework, integrated with Vision-Language Models, demonstrates a strong capability to convert hateful memes into non-hateful forms.
• Score: 12.929357709840975
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: The rapid evolution of social media has provided enhanced communication channels for individuals to create online content, enabling them to express their thoughts and opinions. Multimodal memes, often utilized for playful or humorous expressions with visual and textual elements, are sometimes misused to disseminate hate speech against individuals or groups. While the detection of hateful memes is well-researched, developing effective methods to transform hateful content in memes remains a significant challenge. Leveraging the powerful generation and reasoning capabilities of Vision-Language Models (VLMs), we address the tasks of detecting and mitigating hateful content. This paper presents two key contributions: first, a definition-guided prompting technique for detecting hateful memes, and second, a unified framework for mitigating hateful content in memes, named UnHateMeme, which works by replacing hateful textual and/or visual components. With our definition-guided prompts, VLMs achieve impressive performance on hateful memes detection task. Furthermore, our UnHateMeme framework, integrated with VLMs, demonstrates a strong capability to convert hateful memes into non-hateful forms that meet human-level criteria for hate speech and maintain multimodal coherence between image and text. Through empirical experiments, we show the effectiveness of state-of-the-art pretrained VLMs such as LLaVA, Gemini and GPT-4o on the proposed tasks, providing a comprehensive analysis of their respective strengths and limitations for these tasks. This paper aims to shed light on important applications of VLMs for ensuring safe and respectful online environments.

Related papers

• Improving Multimodal Hateful Meme Detection Exploiting LMM-Generated Knowledge [11.801596051153725]
  detecting hateful content in memes has emerged as a task of critical importance.<n>We propose to address the task leveraging knowledge encoded in powerful Large Multimodal Models (LMM)<n>Specifically, we propose to exploit LMMs in a two-fold manner. First, by extracting knowledge oriented to the hateful meme detection task in order to build strong meme representations.
  arXiv  Detail & Related papers  (2025-04-14T06:23:44Z)
• HateSieve: A Contrastive Learning Framework for Detecting and Segmenting Hateful Content in Multimodal Memes [8.97062933976566]
  textscHateSieve is a framework designed to enhance the detection and segmentation of hateful elements in memes.<n>textscHateSieve features a novel Contrastive Meme Generator that creates semantically paired memes.<n> Empirical experiments on the Hateful Meme show that textscHateSieve not only surpasses existing LMMs in performance with fewer trainable parameters but also offers a robust mechanism for precisely identifying and isolating hateful content.
  arXiv  Detail & Related papers  (2024-08-11T14:56:06Z)
• White-box Multimodal Jailbreaks Against Large Vision-Language Models [61.97578116584653]
  We propose a more comprehensive strategy that jointly attacks both text and image modalities to exploit a broader spectrum of vulnerability within Large Vision-Language Models. Our attack method begins by optimizing an adversarial image prefix from random noise to generate diverse harmful responses in the absence of text input. An adversarial text suffix is integrated and co-optimized with the adversarial image prefix to maximize the probability of eliciting affirmative responses to various harmful instructions.
  arXiv  Detail & Related papers  (2024-05-28T07:13:30Z)
• Exploring the Limits of Zero Shot Vision Language Models for Hate Meme Detection: The Vulnerabilities and their Interpretations [9.970031080934003]
  We study the effectiveness of modern day vision language models (VLMs) in handling intricate tasks such as hate meme detection.<n>We perform thorough prompt engineering and query state-of-the-art VLMs using various prompt types to detect hateful/harmful memes.
  arXiv  Detail & Related papers  (2024-02-19T15:03:04Z)
• Meme-ingful Analysis: Enhanced Understanding of Cyberbullying in Memes Through Multimodal Explanations [48.82168723932981]
  We introduce em MultiBully-Ex, the first benchmark dataset for multimodal explanation from code-mixed cyberbullying memes. A Contrastive Language-Image Pretraining (CLIP) approach has been proposed for visual and textual explanation of a meme.
  arXiv  Detail & Related papers  (2024-01-18T11:24:30Z)
• Mapping Memes to Words for Multimodal Hateful Meme Classification [26.101116761577796]
  Some memes take a malicious turn, promoting hateful content and perpetuating discrimination. We propose a novel approach named ISSUES for multimodal hateful meme classification. Our method achieves state-of-the-art results on the Hateful Memes Challenge and HarMeme datasets.
  arXiv  Detail & Related papers  (2023-10-12T14:38:52Z)
• DisinfoMeme: A Multimodal Dataset for Detecting Meme Intentionally Spreading Out Disinformation [72.18912216025029]
  We present DisinfoMeme to help detect disinformation memes. The dataset contains memes mined from Reddit covering three current topics: the COVID-19 pandemic, the Black Lives Matter movement, and veganism/vegetarianism.
  arXiv  Detail & Related papers  (2022-05-25T09:54:59Z)
• DISARM: Detecting the Victims Targeted by Harmful Memes [49.12165815990115]
  DISARM is a framework that uses named entity recognition and person identification to detect harmful memes. We show that DISARM significantly outperforms ten unimodal and multimodal systems. It can reduce the relative error rate for harmful target identification by up to 9 points absolute over several strong multimodal rivals.
  arXiv  Detail & Related papers  (2022-05-11T19:14:26Z)
• Detecting and Understanding Harmful Memes: A Survey [48.135415967633676]
  We offer a comprehensive survey with a focus on harmful memes. One interesting finding is that many types of harmful memes are not really studied, e.g., such featuring self-harm and extremism. Another observation is that memes can propagate globally through repackaging in different languages and that they can also be multilingual.
  arXiv  Detail & Related papers  (2022-05-09T13:43:27Z)
• Memes in the Wild: Assessing the Generalizability of the Hateful Memes Challenge Dataset [47.65948529524281]
  We collect hateful and non-hateful memes from Pinterest to evaluate out-of-sample performance on models pre-trained on the Facebook dataset. We find that memes in the wild differ in two key aspects: 1) Captions must be extracted via OCR, and 2) Memes are more diverse than traditional memes', including screenshots of conversations or text on a plain background.
  arXiv  Detail & Related papers  (2021-07-09T09:04:05Z)
• Multimodal Learning for Hateful Memes Detection [6.6881085567421605]
  We propose a novel method that incorporates the image captioning process into the memes detection process. Our model achieves promising results on the Hateful Memes Detection Challenge.
  arXiv  Detail & Related papers  (2020-11-25T16:49:15Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.