From Native Memes to Global Moderation: Cross-Cultural Evaluation of Vision-Language Models for Hateful Meme Detection

• URL: http://arxiv.org/abs/2602.07497v2
• Date: Wed, 11 Feb 2026 22:19:33 GMT
• Title: From Native Memes to Global Moderation: Cross-Cultural Evaluation of Vision-Language Models for Hateful Meme Detection
• Authors: Mo Wang, Kaixuan Ren, Pratik Jalan, Ahmed Ashraf, Tuong Vy Vu, Rahul Seetharaman, Shah Nawaz, Usman Naseem,
• Abstract summary: We introduce a systematic evaluation framework designed to quantify the cross-cultural robustness of state-of-the-art vision-language models (VLMs)<n>We analyze three axes: (i) learning strategy (zero-shot vs. one-shot), (ii) prompting language (native vs. English), and (iii) translation effects on meaning and detection.<n>Results show that the common translate-then-detect'' approach deteriorates performance, while culturally aligned interventions - native-language prompting and one-shot learning - significantly enhance detection.
• Score: 13.900106805972
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Cultural context profoundly shapes how people interpret online content, yet vision-language models (VLMs) remain predominantly trained through Western or English-centric lenses. This limits their fairness and cross-cultural robustness in tasks like hateful meme detection. We introduce a systematic evaluation framework designed to diagnose and quantify the cross-cultural robustness of state-of-the-art VLMs across multilingual meme datasets, analyzing three axes: (i) learning strategy (zero-shot vs. one-shot), (ii) prompting language (native vs. English), and (iii) translation effects on meaning and detection. Results show that the common ``translate-then-detect'' approach deteriorate performance, while culturally aligned interventions - native-language prompting and one-shot learning - significantly enhance detection. Our findings reveal systematic convergence toward Western safety norms and provide actionable strategies to mitigate such bias, guiding the design of globally robust multimodal moderation systems.

Related papers

• When Meanings Meet: Investigating the Emergence and Quality of Shared Concept Spaces during Multilingual Language Model Training [57.230355403478995]
  We investigate the development of language-agnostic concept spaces during pretraining of EuroLLM.<n>We find that shared concept spaces emerge early and continue to refine, but that alignment with them is language-dependent.<n>In contrast to prior work, our fine-grained manual analysis reveals that some apparent gains in translation quality reflect shifts in behavior.
  arXiv  Detail & Related papers  (2026-01-30T11:23:01Z)
• I Am Aligned, But With Whom? MENA Values Benchmark for Evaluating Cultural Alignment and Multilingual Bias in LLMs [5.060243371992739]
  We introduce MENAValues, a novel benchmark designed to evaluate the cultural alignment and multilingual biases of large language models (LLMs)<n> Drawing from large-scale, authoritative human surveys, we curate a structured dataset that captures the sociocultural landscape of MENA with population-level response distributions from 16 countries.<n>Our analysis reveals three critical phenomena: "Cross-Lingual Value Shifts" where identical questions yield drastically different responses based on language, "Reasoning-Induced Degradation" where prompting models to explain their reasoning worsens cultural alignment, and "Logit Leakage" where models refuse sensitive questions while internal probabilities reveal strong hidden
  arXiv  Detail & Related papers  (2025-10-15T05:10:57Z)
• BLEnD-Vis: Benchmarking Multimodal Cultural Understanding in Vision Language Models [54.16874020794336]
  We introduce BLEnD-Vis, a benchmark designed to evaluate the robustness of everyday cultural knowledge in vision-language models (VLMs)<n> BLEnD-Vis constructs 313 culturally grounded question templates spanning 16 regions and generates three aligned multiple-choice formats.<n>The resulting benchmark comprises 4,916 images and over 21,000 multiple-choice question (MCQ) instances, validated through human annotation.
  arXiv  Detail & Related papers  (2025-10-13T09:10:05Z)
• MMA-ASIA: A Multilingual and Multimodal Alignment Framework for Culturally-Grounded Evaluation [91.22008265721952]
  MMA-ASIA centers on a human-curated, multilingual, and multimodally aligned benchmark covering 8 Asian countries and 10 languages.<n>This is the first dataset aligned at the input level across three modalities: text, image (visual question answering), and speech.<n>We propose a five-dimensional evaluation protocol that measures: (i) cultural-awareness disparities across countries, (ii) cross-lingual consistency, (iii) cross-modal consistency, (iv) cultural knowledge generalization, and (v) grounding validity.
  arXiv  Detail & Related papers  (2025-10-07T14:12:12Z)
• Multilingual Vision-Language Models, A Survey [0.9722250595763385]
  This survey examines multilingual vision-language models that process text and images across languages.<n>We review 31 models and 21 benchmarks, spanning encoder-only and generative architectures.
  arXiv  Detail & Related papers  (2025-09-26T09:46:13Z)
• SESGO: Spanish Evaluation of Stereotypical Generative Outputs [1.1549572298362782]
  This paper addresses the critical gap in evaluating bias in multilingual Large Language Models (LLMs)<n>Current evaluations remain predominantly US-English-centric, leaving potential harms in other linguistic and cultural contexts largely underexamined.<n>We introduce a novel, culturally-grounded framework for detecting social biases in instruction-tuned LLMs.
  arXiv  Detail & Related papers  (2025-09-03T14:04:51Z)
• From Word to World: Evaluate and Mitigate Culture Bias in LLMs via Word Association Test [50.51344198689069]
  We extend the human-centered word association test (WAT) to assess the alignment of large language models with cross-cultural cognition.<n>To address culture preference, we propose CultureSteer, an innovative approach by embedding cultural-specific semantic associations directly within the model's internal representation space.
  arXiv  Detail & Related papers  (2025-05-24T07:05:10Z)
• RAVENEA: A Benchmark for Multimodal Retrieval-Augmented Visual Culture Understanding [79.44246283490665]
  We introduce RAVENEA, a new benchmark designed to advance visual culture understanding through retrieval.<n>RAVENEA focuses on two tasks: culture-focused visual question answering (cVQA) and culture-informed image captioning (cIC)<n>We train and evaluate seven multimodal retrievers for each image query, and measure the downstream impact of retrieval-augmented inputs across fourteen state-of-the-art vision-language models.
  arXiv  Detail & Related papers  (2025-05-20T14:57:16Z)
• JiraiBench: A Bilingual Benchmark for Evaluating Large Language Models' Detection of Human Self-Destructive Behavior Content in Jirai Community [9.492476871323763]
  This paper introduces JiraiBench, the first bilingual benchmark for evaluating large language models' effectiveness in detecting self-destructive content.<n>We focus on the transnational "Jirai" (landmine) online subculture that encompasses multiple forms of self-destructive behaviors including drug overdose, eating disorders, and self-harm.<n>Our dataset comprises 10,419 Chinese posts and 5,000 Japanese posts with multidimensional annotation along three behavioral categories.
  arXiv  Detail & Related papers  (2025-03-27T16:48:58Z)
• Benchmarking Machine Translation with Cultural Awareness [50.183458829028226]
  Translating culture-related content is vital for effective cross-cultural communication. Many culture-specific items (CSIs) often lack viable translations across languages. This difficulty hinders the analysis of cultural awareness of machine translation systems.
  arXiv  Detail & Related papers  (2023-05-23T17:56:33Z)

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.