The authenticity of Hadiths, or sayings and actions attributed to the Prophet Muhammad peace be upon him, is a critical issue in Islamic scholarship. In this paper, we present a method for classifying Prophet’s statements as authentic or unauthentic using artificial intelligence (AI) techniques in our newly created corpus. Our corpus is composed of a collection of wordings separated from the text (Maten) and chain of narrators and, then, labelled based on their authenticity. In the first step, we will focus on analyzing the style of the Prophet Muhammad’s wordings and evaluating the performance of machine learning (ML) techniques for the classification of hadith authenticity. Second, we use a BERT-based model, which is a ML framework for natural language processing (NLP) designed to capture the contextual relationships between the words in the hadith. The results of our experiments show that the hybrid approach is able to achieve high accuracy in authenticating Hadiths, outperforming machine learning based methods.