“Helpful” never outranks “verified.”
Our working constitution for source selection, AI boundaries, moderation, and corrections.
The non-negotiable rule
Quraan Journey may display religious content only when it comes from an identified approved source record. If a claim cannot be cited precisely and reviewed at the required level, it remains unpublished.
- 01
Qur’an text is immutable application data
The Arabic text is imported verbatim from Tanzil Uthmani 1.1. Installation verifies the downloaded file’s SHA-256 hash and requires exactly 6,236 records. The application never asks a language model to supply, complete, spell-correct, or translate an ayah.
- 02
Metadata is labeled as metadata
Surah names, ayah counts, page boundaries, classifications, and revelation order use Tanzil Quran Metadata 1.0. They are not presented as new commentary.
- 03
AI is a constrained router
The guide first retrieves candidates from approved application data. If AI routing is enabled, the model may return only a structured intent, supplied candidate IDs, and a scholar-required flag. The server rejects IDs that were not supplied. Fixed application templates—not model prose—present results.
- 04
Scholar-required requests stop
Fatwas, creed disputes, hadith grading, personal rulings, and interpretive conclusions are outside the guide. A boundary response directs the visitor to a qualified scholar and does not improvise.
- 05
Community speech is unpublished by default
Submissions enter a private pending queue. Approval must be explicit. Unsupported religious claims, abuse, sectarian attacks, or copied material without rights do not publish.
- 06
Corrections are visible and conservative
A reported issue is first hidden or source-paused when risk warrants it, then investigated against the named edition. Qur’an text updates are accepted only through the upstream verified source and a new integrity record.
Required gates for tafsir, hadith, translations, and stories
Found a concern?
Send the exact URL, visible wording, and reason for concern. High-risk content can be paused while it is checked.