Scholars draw the line on AI in fatwa: human mufti stays in the loop

As bodies from al-Azhar and the International Islamic Fiqh Academy to the Assembly of Muslim Jurists of America build their 2026 programmes around artificial intelligence and ifta, a shared position is emerging — and it is a governance position, not a theological panic. The starting premise is that a large language model can retrieve, compare and draft, but it cannot satisfy the classical conditions of a mufti: it does not bear legal responsibility, cannot weigh a questioner's circumstances, and cannot distinguish quoting an existing ruling (naql) from deriving a new one (istinbat).

From that premise the argument moves to design rather than prohibition. The proposal on the table is a Shariah Governance Framework for AI integration with concrete controls: a human mufti kept in the loop at the final stage of any ruling; regular ethical and bias audits of the models; mandatory disclosure whenever a tool has been used; a national AI-in-Ifta oversight committee sitting under a central Shariah board; and continuing AI-literacy training for advisors. The gains it accepts are real but ancillary — faster retrieval of classical fatwas, easier comparative fiqh, quicker preliminary drafting, more efficient Shariah audit. The line it draws is that assistance is welcome while authority stays human. It is a template other traditions wrestling with automated expertise could study: adopt the tooling, keep the accountability where it belongs.

This is a QeRN summary by Ahmed Qerni. Read the original at Int'l Journal of Islamic & Interdisciplinary Studies: https://ijiissh.com/v3i1/12.php.