The proliferation of fake information on digital and social media platforms have become a surging concern for society. With the growing reliance on the internet as a source of information, it has become increasingly crucial to detect and mitigate the spread of fake news. To address this challenge, the field of Natural Language Processing and Machine Learning has directed considerable efforts toward the development of effective Fake News Identification (FNI) methods. However, the lack of comprehensive and balanced datasets for fake news (FN) detection in a low-resourced language such as the Arabic language remains a major obstacle in this field. To bridge this gap, this research proposes an effective taxonomy for Arabic FN datasets. The taxonomy provides insight of the characteristics and specifications of Arabic FN. The taxonomy is based on an extensive analysis of existing Arabic datasets and relevant literature in the field. This taxonomy can provide a useful framework for the building, categorization, and comparison of FN in the Arabic language and offer a clear understanding of the different types of fake news and how they can be differentiated. Furthermore, this taxonomy provides solid ground for the development of high-quality and balanced Arabic datasets that can effectively facilitate the development of FNI models in the Arabic language. In conclusion, this research paper offers a valuable contribution to the field of FNI by proposing an effective taxonomy for Arabic fake news datasets and support for building comprehensive and balanced datasets in the Arabic language.