Halal certification: When a system errs, who owns up

In May 2025, jurists from across the Muslim world gathered in Doha, Qatar, for the 26th session of the International Islamic Fiqh Academy, the OIC (Organisation of Islamic Cooperation) body for collective legal reasoning. During one of its scientific sessions dedicated to artificial intelligence (AI), 33 research papers were reviewed. 

The resulting text, Resolution 258, is brief. It holds that developing and using AI is permissible by default, subject to six conditions:

1. Its purposes and outcomes must be lawful;
2. It must bring benefit and prevent harm;
3. It must respect beliefs;
4. Protect rights and data;
5. Preserve security;
6. It must adhere to honesty, proper documentation and transparency. 

The Academy recommended a further study on whether AI could be granted legal personality. That recommendation leaves the central question for halal certifiers open. If a machine cannot be a legal person, responsibility for its errors must rest with a human or an institution. Classical jurisprudence has detailed tools for deciding which.