Can Thinking Models Think to Detect Hateful Memes?

Mohamed Bayan Kmainasi, Mucahid Kutlu, Ali Ezzat Shahroor +2

Hateful memes often require compositional multimodal reasoning: the image and text may appear benign in isolation, yet their interaction conveys harmful intent. Although thinking-b…