摘要：ArGuard 是一项关于阿拉伯语迷因和 LLM 提示词中有害内容检测的共享任务。该任务包含两条赛道：赛道 A 聚焦阿拉伯语迷因中的多模态仇恨检测，赛道 B 则针对阿拉伯语 LLM 安全评估中的有害提示词检测。共有 58 支团队注册，35 支团队参与最终评测，27 支团队提交了系统描述论文。参赛团队探索了 AraBERT、Jais 和 Qwen3-VL 等模型。最佳系统在 A1、A2、B1 和 B2 上分别取得了 0.823、0.419、0.984 和 0.790 的宏平均 F1 分数。A2 中的细粒度迷因分类是最具挑战性的设置，部分原因是标签稀疏，以及训练集与测试集之间存在分布偏移。

Abstract:ArGuard is a shared task on harmful content detection in Arabic memes and LLM prompts. It includes two tracks: Track A focuses on multimodal hate detection in Arabic memes, while Track B addresses harmful prompt detection for Arabic LLM safety evaluation. In total, 58 teams registered, 35 participated in the final evaluation, and 27 submitted system-description papers. Participating teams explored models such as AraBERT, Jais, and Qwen3-VL. The best systems achieved macro-F1 scores of 0.823 on A1, 0.419 on A2, 0.984 on B1, and 0.790 on B2. Fine-grained meme classification in A2 was the most challenging setting, partly due to sparse labels and train-test distribution shifts.