Read our AI Policy for research and editorial integrity standards.

AI Policy

Madain Project’s AI Policy — Editorial and Research Integrity Framework defines strict guidelines for the responsible use of artificial intelligence in research, writing, and publication workflows. It ensures all content is human-authored, AI-assisted only within limited editorial boundaries, fully disclosed, and independently verified against original sources.
Use Images
  • Core Principle
    Madain Project treats AI as a limited assistive tool for language and workflow support only. AI does not participate in research, does not function as a source of knowledge, and does not contribute to historical, archaeological, theological, or interpretive conclusions. All intellectual responsibility remains with human editors and reviewers. Human authorship is mandatory. AI cannot originate publishable content, and human editors must always produce the foundational intellectual material. The final publication draft must be at least 95% or more human written.

  • Permitted Use of AI
    AI may be used only for constrained editorial assistance, primarily in language refinement and structural clarity. This includes sentence restructuring, grammatical correction, stylistic improvement, and limited rephrasing of existing human-written content. Editors may also use AI in narrowly defined discovery tasks where external sources are later independently retrieved and verified by humans. Any such use must be disclosed within the internal editorial workflow. AI may assist in image description workflows under strict conditions. It may identify visually observable elements and provide preliminary descriptive suggestions. However, any historical, archaeological, linguistic, chronological, or interpretive content derived from images must be independently verified by human editors before publication.

  • Prohibited Use of AI
    AI must never be used to generate complete articles, standalone paragraphs, original or even secondary research narratives. It cannot create or replace human-authored academic or historical interpretation. AI cannot be used for summarizing academic or primary sources, including journal articles or restricted databases. It cannot be used to interpret difficult scholarly passages, evaluate historical claims, or determine scholarly correctness. AI must not be used as a source of citations or references. All citations must be independently located, read, and verified by human editors. No unpublished, confidential, licensed, subscription-based, or restricted material may be submitted to any external AI system under any circumstances. AI is also prohibited from participating in archaeological interpretation, historical reconstruction, or resolution of competing scholarly perspectives.

  • Editorial Standards and Historical Interpretation
    Madain Project requires that all major scholarly perspectives be presented without editorial bias. Articles must reflect competing interpretations where they exist, without forcing a definitive conclusion in the body of the text.
    AI may not propose historical conclusions, reconcile conflicting scholarly positions, or determine correctness among competing academic views. It may only assist in language refinement of already established interpretations. Editorial notes, when used, may contain personal observations by human editors but must remain separate from the main article body.

  • Image Description and Experimental Features
    AI-assisted image description is permitted in experimental workflows. AI may describe visible features of images and suggest limited contextual possibilities, but must not assign definitive historical attribution, dating, or linguistic interpretation without human verification. All experimental content generated with AI assistance is clearly labeled on the website as an experimental archive feature. This disclosure informs users that AI has been used in generating and refining descriptions alongside human editorial review.

  • Disclosure and Recordkeeping
    AI usage must be fully documented within the editorial system. Editors must record whether AI was used, the intent of use, the system involved, and the date of use. Where relevant, supporting outputs are attached to the editorial record. Disclosure of AI usage is required for all published content that involved AI assistance. This includes tagging at the claim or paragraph level using internal markers distinguishing AI-assisted content from human-verified content. All editorial records related to AI use are retained permanently within the internal system. Readers may request information about AI involvement in specific published material, and such disclosures must be made available in accordance with internal records.

  • Authorship and Attribution
    AI is never credited as an author or contributor under any circumstance. Authorship remains exclusively human. AI usage does not alter authorship attribution but must be transparently disclosed through internal tagging and publication-level markers where applicable. Experimental archive pages remain explicitly labeled as AI-assisted, while standard articles follow internal disclosure protocols without requiring prominent public AI acknowledgments unless specified by experimental classification.

  • Classification of AI-Assisted Writing
    AI-assisted writing is defined as any text where AI was used for refinement, restructuring, or clarity improvements on existing human-written content, provided that the editor maintains substantive authorship and the reviewer confirms compliance with transformation thresholds and editorial standards. AI-generated content is defined as any text that originates from AI without a pre-existing human-written draft or substantive human intellectual structure. Such content is strictly prohibited. AI-influenced content refers to cases where AI suggestions materially shape phrasing or structure but do not replace human authorship. These cases remain permitted only under disclosure and reviewer approval, provided that human rewriting meets the minimum transformation threshold. A hybrid editorial system is used in which editors apply a minimum threshold of human rewriting (generally at least twenty percent meaningful modification following AI assistance which again needs to be verified by a reviewing editor), and reviewers validate both compliance and academic integrity against source material.

Let's bring some history to your inbox

Signup for our monthly newsletter / online magazine.
No spam, we promise.

Privacy Policy



Top