A study aid, not a muftī.
Last updated: September 2026
Joyful Muslims uses AI in three places, and nowhere else. This page names each one, says what it is fed, and — most importantly — what none of them is.
Where AI is used
- The "Ask the library" box on this website. On our Ask pages you can type a question and receive an answer built from passages in the books we publish, with citations you can open and read in full. A large language model (Anthropic's Claude) composes the answer from those passages under a system prompt that anchors it in the Qurʾān, the authentic Sunnah, and mainstream Sunni scholarship, and asks it to cite where possible.
- Translation drafting. Our English translations are drafted with the help of Claude and GPT-4 from the original Arabic, then reviewed before they enter the library. Every translated book says so at its head, before it begins. See how we translate and narrate.
- Narration. The audiobooks are read by a synthesized voice produced under our documented method — Islamic honorifics spoken in full, Arabic names pronounced correctly — and reviewed before release. The Qurʾān is never synthesized: it is the recitation of Shaykh Maḥmūd Khalīl al-Ḥuṣarī.
The app contains no AI chat. Earlier versions carried an in-app assistant called JM Scholar; it was retired in version 3.1 (September 2026). Nothing you type in the app is sent to an AI provider.
What the answer box is not
- Not a muftī. It does not issue fatāwā. It cannot take the place of a qualified scholar who knows you and your circumstances.
- Not a source of legal, medical, psychological, or financial advice. For personal matters, speak to a qualified professional or your local imam.
- Not infallible. Large language models can make mistakes — including misquoting verses, mixing up hadith numbering, and over-confidently summarising views they have compressed. It is asked to cite its sources; you should verify what matters.
- Not a spiritual authority. Reading a book or asking a machine does not make one a scholar. Traditional Islamic knowledge is taught through teachers, not apps.
What is sent, and what is kept
- When you use the answer box, your question text is sent over an encrypted connection to Anthropic. Nothing that identifies you — no name, email, account, or payment detail — goes with it. You do not need an account to use it.
- Questions are not used to train any model: Anthropic and OpenAI do not train on API data by default, and we do not opt in.
- We log the length of questions and model-usage statistics so we can measure quality and cost, not the text itself for training.
See our Privacy Policy for full detail.
When to ask a real scholar instead
Please do not rely on an AI answer for:
- Rulings on personal worship, marriage, divorce, inheritance, custody, business contracts, or any legal matter.
- Questions where your specific situation, intention, or state changes the answer — that is exactly where a trusted scholar who knows you is irreplaceable.
- Sensitive or controversial topics where misunderstanding can cause harm.
For matters like these, contact your local imam, a trusted muftī, or a scholarly organisation in your madhhab.
Mistakes and corrections
If you spot an error — an incorrect citation, a misattributed view, or a mistranslation — we want to know. Email [email protected] with the question and the response, and we will investigate and correct the source material or the system prompt where appropriate.
Summary
AI can make the library easier to enter. It cannot replace scholarship, a teacher, or the barakah of sitting with people of knowledge. Use it as an opening, not a conclusion.