Artificial Intelligence in Islamic Guidance: Assessing Chatbot Performance and Jurisprudential Adherence

Abstract The rise of AI chatbots for Islamic guidance has prompted questions about their accuracy and adherence to Islamic jurisprudence (fiqh) . This study assessed five Islamic chatbots using Bloom’s Taxonomy, Islamic accuracy from the perspectives of Ahl al-ḥadīth and Ahl al-raʾy , and the risks of AI hallucinations. The evaluation involved 80 questions on Islamic law, which were analyzed for cognitive level, clarity, reasoning, textual evidence, and methodological soundness. The results revealed significant performance variability, with the chatbots failing to meet the standards of Islamic legal reasoning. Issues included misattribution of Qurʾānic verses and ḥadīth , oversimplification, and inconsistent responses. No chatbot achieved the “Trusted Educational Output” status, underscoring the epistemic challenges in religious domains, where knowledge depends on authenticated transmission chains ( isnād ). This study recommends collaboration between Islamic scholars and AI developers to create epistemologically grounded systems. Although AI can improve access to Islamic knowledge, it cannot substitute trained Islamic scholars ( Muftīs ).