# FBHM: Functional Benchmarking and Steering of VLMs for Hateful Meme Detection

arXiv: https://arxiv.org/abs/2605.31349 · HTML: https://mlsanity.com/paper/2605.31349/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Social bias and fairness in VLMs, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 17 of its 39 references that are in the map.
- First posted on arXiv 29 May 2026.

## Placed in

- Multimodal LLMs / vision-language models / [Social bias and fairness in VLMs](https://mlsanity.com/mllm/social-bias-and-fairness-in-vlms/)

## Builds on (most cited first, 17 of 17)

- [QLoRA: Efficient Finetuning of Quantized LLMs](https://mlsanity.com/paper/2305.14314/) (2023)
- [Training Compute-Optimal Large Language Models](https://mlsanity.com/paper/2203.15556/) (2022)
- [Beyond Accuracy: Behavioral Testing of NLP models with CheckList](https://mlsanity.com/paper/2005.04118/) (2020)
- [Winoground: Probing Vision and Language Models for Visio-Linguistic Compositionality](https://mlsanity.com/paper/2204.03162/) (2022)
- [The Hateful Memes Challenge: Detecting Hate Speech in Multimodal Memes](https://mlsanity.com/paper/2005.04790/) (2020)
- [In-context Vectors: Making In Context Learning More Effective and Controllable Through Latent Space Steering](https://mlsanity.com/paper/2311.06668/) (2023)
- [HateCheck: Functional Tests for Hate Speech Detection Models](https://mlsanity.com/paper/2012.15606/) (2020)
- [Detecting Harmful Memes and Their Targets](https://mlsanity.com/paper/2110.00413/) (2021)
- [GOAT-Bench: Safety Insights to Large Multimodal Models through Meme-Based Social Abuse](https://mlsanity.com/paper/2401.01523/) (2024)
- [Decoding the Underlying Meaning of Multimodal Hateful Memes](https://mlsanity.com/paper/2305.17678/) (2023)
- [MemeGuard: An LLM and VLM-based Framework for Advancing Content Moderation via Meme Intervention](https://mlsanity.com/paper/2406.05344/) (2024)
- [Demystifying Hateful Content: Leveraging Large Multimodal Models for Hateful Meme Detection with Explainable Decisions](https://mlsanity.com/paper/2502.11073/) (2025)
- [Exploring the Limits of Zero Shot Vision Language Models for Hate Meme Detection: The Vulnerabilities and their Interpretations](https://mlsanity.com/paper/2402.12198/) (2024)
- [Read as You See: Guiding Unimodal LLMs for Low-Resource Explainable Harmful Meme Detection](https://mlsanity.com/paper/2506.08477/) (2025)
- [PEST: Parameter Efficient Steering of Blackbox VLMs via Agentic Few-shot Alignment for Hateful Meme Moderation](https://mlsanity.com/paper/2601.04692/) (2026)
- [MemeSense: An Adaptive In-Context Framework for Social Commonsense Driven Meme Moderation](https://mlsanity.com/paper/2502.11246/) (2025)
- [The Enforcement and Feasibility of Hate Speech Moderation](https://mlsanity.com/paper/2604.12289/) (2026)

## Abstract (arXiv)

Hateful meme detection remains a formidable challenge for vision-language models, as existing benchmarks are structurally observational - confounding rhetorical hate mechanisms with target community features and preventing causal evaluation of model vulnerabilities. To address this, we introduce FBHM, a systematically curated benchmark of Functionality Based Hateful Memes constructed along two orthogonal axes: 25 distinct rhetorical functionalities and 10 target communities (5,000 memes total). Benchmarking state-of-the-art VLMs reveals a severe generalization gap: models highly accurate on standard datasets catastrophically drop to near-random performance on FBHM, proving they exploit dataset-specific heuristics rather than robust multimodal reasoning. To efficiently close this gap, we propose LSV (learnable steering vectors), an ultra-low data regime strategy that applies a causal intervention objective on as few as 500 steering samples (50 unique base memes), boosting FBHM performance by ~30 Macro-F1 points while outperforming in-context learning and PEFT without degrading source-domain performance.