The goal of news image captioning is to generate captions by integrating news article content with corresponding images, highlighting the relationship between textual context and visual elements. The majority of research on news image captioning focuses on English, primarily because datasets in other languages are scarce. To address this limitation, we create the first large-scale multilingual news image captioning benchmark, LaMuN, comprising more than 800,000 news images in 59 languages, including several low-resource languages. We further propose a novel chain-of-thought method for automatic news image captioning that achieves state-of-the-art results across a diverse set of languages in LaMuN. Together, the benchmark and the proposed method open new avenues for further advances in developing and evaluating multilingual multimodal large language models.