GuardHarMem: A Multimodal Dataset for Harmful Meme Classification

Harmful content on social media, especially in the form of memes, poses unique challenges for moderation and analysis. Memes combine text and imagery to convey complex messages rapidly and evoke emotional responses, enabling the dissemination of harmful ideas, reinforcement of stereotypes, and normalization of discriminatory behaviour, often eluding standard moderation tools. Existing datasets, such as Hateful Memes (\≈10 K samples) and Memotion (\≈8 K) focus on binary or coarse-grained labels and omit many common forms of harm. To address this gap, we introduce GuardHarMem, a new corpus of \≈16.6 K memes annotated with fine-grained categories including racism, mockery, and promotion of harmful substances.