<![CDATA[Computational research on the Qurʾan remains centred on the sound and script of the text, while the learning of meaning (the domain of tafsīr) has attracted far less attention. This review maps the peer-reviewed evidence on artificial intelligence (AI)-based models for learning Tafsīr. Following the Arksey and OʾMalley framework and reporting against PRISMA-ScR, 240 Scopus-indexed records (2015–2026) were screened in a two-stage, dual-pass process. Eligible studies combined an AI component, a Qurʾanic-meaning object, and a learning component; nine were included and charted against five research questions. Eight of nine studies appeared in 2025–2026 and seven originated in Southeast and South Asia. Across three technology clusters, they divide into learning-tested studies (n = 2), learner-facing systems (n = 5), and technical systems with learning relevance (n = 2). Only the first group measured learning outcomes, neither with a control group; the rest reported system, usability, content-quality, or perception results. Concerns about absent sanad, unreliable citation, and mishandled classical Arabic recurred across independent studies. The review searched a single database in English only; 11 of 37 reports sought were unretrievable; and no critical appraisal was undertaken. The findings map an evidence base rather than warrant effectiveness claims or support deployment decisions. Within the Scopus-indexed English-language literature, this appears to be the first review isolating meaning-oriented AI in Qurʾanic education. It identifies a technical–pedagogical disconnect and argues that interpretive authority is an emerging design requirement indicated by the mapped literature, proposing an agenda uniting source-grounded architectures, explicit tafsīr pedagogy, and sanad-aware governance.]]>
