Imagine it is almost midnight and a question about prayer is nagging at you. The scholar you usually ask is asleep. So you open a chat window, type your question, and an answer appears in seconds. In Dubai, that stopped being imaginary on 29 October 2019. The Islamic Affairs and Charitable Activities Department (IACAD) launched "Virtual Ifta," described at the time as the world's first service to use artificial intelligence to issue fatwas.

It was a big moment, and it left behind an even bigger question. Not "can a machine talk about religion?" It clearly can. The real question is this: when it does, who is answerable? That is the story I want to tell here.

A Small Experiment with Big Ambitions

Virtual Ifta started modestly. It could answer a little over 200 questions about prayer, in Arabic and English, through the IACAD website. The team planned to add WhatsApp and more topics later, such as fasting and finance. It belonged to Dubai 10X, a government programme meant to keep the city ten years ahead in public services.

Why build it at all? Because of the numbers. IACAD's Grand Mufti said the department receives roughly 130,000 fatwa requests every year. Questions are many, and scholars are few.

What I find most interesting is the caution built into it. Every user was told that the answers were generated by AI and that mistakes were possible. The Grand Mufti also said that if the assistant was not at least 85 per cent sure what you were asking, it would not answer. It offered similar questions or asked you to rephrase. A machine that knows when to stay quiet is rarer than you might think.

Cairo noticed too. A scholarly study of the two cases reports that Al-Azhar's Fatwa Global Centre began its own virtual ifta service in January 2020, only months after Dubai.

Cairo Says: Slow Down

Egypt's Dar al-Ifta, one of the oldest fatwa institutions in the Muslim world, took a firm stand. A believer had written that he often asked AI applications for Islamic rulings and wanted to know if that was acceptable. The answer was no: relying on AI applications to obtain a fatwa is impermissible, because they remain outside the scope of fatwa, even where they match human experts at certain tasks.

Notice what the ruling did not say. It did not condemn technology. AI used for a lawful purpose takes the ruling of that purpose, following the maxim that means take the ruling of their ends. The problem was never the tool. The problem was treating its output as a fatwa.

Egypt's Grand Mufti made the point more personally in a later article. He begins with a teenager whose parents keep warning against leaning too much on AI, and who asks: would the Prophet (peace be upon him) use artificial intelligence if he were alive today? His answer is balanced. Producing an answer is not the same as having the authority to give a fatwa, and AI does not carry a mufti's responsibility. Yet institutions can use it well to organise knowledge, improve access and reach new generations. In his words, it should be a servant of knowledge, never its master.

When the Question Was the Quran Itself

Early in 2026, news outlets reported another ruling from Dar al-Ifta. This time the subject was scripture. Using AI tools such as ChatGPT to interpret the Quran was declared impermissible, because depending entirely on their output exposes the sacred text to guesswork, inaccuracy and misrepresentation.

The reasons were down to earth. These systems cannot verify their sources. No specialist oversees them. And they have no established scientific method of interpretation.

Again, this was not a rejection of the digital world. Back in June 2022, Dar al-Ifta had launched FatwaPro, a smartphone app that answers religious queries from Muslims across the globe. The line it drew was clear: technology for access and communication, yes; handing over interpretive authority, no.

Two Cities, One Principle

It is easy to picture Dubai as the enthusiast and Cairo as the sceptic. But look at where Dubai stands today. In March 2026, IACAD described AI as a support for its fatwa work, running inside a multi-stage system designed to protect accuracy at a time when unregulated fatwas spread freely on social media. It has also adopted an Artificial Intelligence and Data Strategy for 2025–2027.

So both institutions are drawing the same line: AI as a helper is one thing, AI as the source of religious authority is another. They only differ on where exactly the line sits. Dubai tested automated answers in a narrow, tightly controlled space. Cairo insisted that a fatwa is, in the end, a human act of responsibility.

Why This Matters

This is exactly the question my own research paper, presented at the 3rd International Conference on Islamic Law (ICIsLaw) 2026, tries to answer. My central claim is simple: AI is an assisting instrument, not an independent source of Shariah authority. Three reasons stand behind it.

First, fluency is not the same as knowledge. A generative system can blend opinions from different schools without their conditions, treat a general principle as a universal rule, or even produce a citation that looks perfect but does not exist. When the subject is worship, family life or finance, that is not a small slip.

Second, the Quran already gave us a method. Ask the people of knowledge if you do not know (16:43). Do not follow what you do not know (17:36). Verify news before acting on it (49:6). A confident answer from a machine is exactly the kind of claim these verses tell us to check.

Third, responsibility needs an address. In a hadith recorded by Abu Dawud (3657), the sin of an unfounded fatwa falls on the one who issued it. A machine cannot carry that burden. If no identifiable person stands behind an answer, accountability simply vanishes.

What a Safer System Looks Like

My paper does not say "ban it." Dubai's safeguards and Cairo's warnings together point to a middle road, which I describe as a seven-stage, human-accountable framework:

1. Understand the technology. What data does it use, how does it retrieve it, and where does it fall short?
2. Define the actual case. Separate the real-life facts from the "AI" label.
3. Classify the legal question. Worship, transactions, family, privacy, and so on.
4. Retrieve and verify sources. AI may search, but every citation is checked against the original.
5. Apply usul al-fiqh and maqasid. Weigh the evidence, the benefit and the harm.
6. Consult technical experts. Cybersecurity, medicine, economics, engineering, as the case needs.
7. Human review and accountability. A named, qualified person who really examined the reasoning.

The last stage matters most, and it must be real. A scholar's name stamped on automated text is not review. The paper also suggests three layers for any religious chatbot: information (verified sources), analysis (cited scholarly positions) and fatwa (a contextual judgment that stays in human hands). Dubai's confidence threshold and Cairo's insistence on scholars both fit this picture.

Privacy belongs in the conversation too. People type family quarrels, health worries and private struggles into religious questions. If a service stores or reuses that information carelessly, it creates a brand-new harm, and the principle of avoiding harm applies to it just as much as to any ruling.

What the Story Teaches

So here is what actually happened. In 2019, a government fatwa department launched an AI service with clear limits built in, and a leading scholarly centre followed within months. Egypt's Dar al-Ifta later ruled that AI cannot be relied on for fatwa or for Quranic interpretation, while still welcoming digital tools for access. And by 2026, Dubai was describing AI as support for its fatwa system, not a replacement for scholars.

The lesson is not that the technology failed, or that tradition refused to change. Serious institutions on both sides arrived at the same idea, which I would sum up like this:

Usefulness is not authority; fluency is not proof; and speed is not a substitute for judgment.

The next decade will not be decided by whether machines can speak about religion. They already do. It will be decided by whether the people who build, use, review and govern them can keep that speech tied to evidence, wisdom and responsibility.

References:

1. Dar al-Ifta al-Misriyyah. "Using AI Applications to Obtain Fatwas." dar-alifta.org/en/fatwa/details/22255
2. Egypt's Grand Mufti. "Resolving Family Issues: The Dilemma of AI-Generated Fatwas." dar-alifta.org/en/article/details/12325
3. Gulf News. "Dubai launches 'world's first' Artificial Intelligence fatwa service," 29 October 2019.
4. Al-Araby Al-Jadeed (English). "Virtual mufti: Dubai launches 'first ever' AI-powered fatwa service."
5. Islamic Affairs and Charitable Activities Department, Dubai. "Artificial Intelligence Supports the Fatwa System in Dubai," 7 March 2026. iacad.gov.ae
6. "Artificial Intelligence on Sunni Islam's Fatwa Issuance in Dubai and Egypt." is.urd.ac.ir/article_161049
7. Middle East AI News; MEA Tech Watch (January 2026). Reports on Dar al-Ifta's ruling on AI-based Quran interpretation.
8. Raza, Asif. "Artificial Intelligence and Islamic Legal Reasoning: Rethinking Fatwa Methodology in the Digital Age." 3rd International Conference on Islamic Law (ICIsLaw), 2026.
9. The Quran, 16:43, 17:36, 49:6. Sunan Abi Dawud, hadith 3657.

.    .    .