This page states how authors, reviewers and editors of QAIRU Journal may use generative AI, what must be disclosed and in what form, and what the journal does when an undisclosed or prohibited use is found.
QAIRU Journal publishes research on artificial intelligence, and the authors it publishes build and study the tools this policy concerns. The policy is therefore written to be used, not to deter: it says what to declare, in what words, and where the line runs between a writing aid and a failure of research integrity.
A generative AI tool cannot be an author. Authorship carries accountability — for the integrity of the work, for responding to questions after publication, and for approving what is submitted — and only a person can hold it. A tool cannot agree to be accountable, cannot consent to publication and cannot answer for an error.
The authors are fully responsible for every part of the submitted work: the text, the code, the figures, the data and the references, including anything produced with a tool. Responsibility does not transfer to the tool, to its provider, or to the fact that a tool made the mistake. See Authorship and Contributions.
Disclose any use of generative AI in preparing the manuscript. The disclosure goes in a named section placed immediately before the references, with the heading Declaration of Generative AI Use, and states:
Model wording to adapt:
Declaration of Generative AI Use. During the preparation of this manuscript the authors used [tool, version] in [month, year] in order to [purpose, for example: improve the readability of the Introduction and Discussion]. After using this tool the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.
Where no generative AI was used, the section may be omitted, or may state: "No generative AI tools were used in the preparation of this manuscript."
Uses that must be disclosed include:
Routine assistive tools do not require a declaration:
This boundary exists so that the declaration keeps its meaning. A declaration that covers every spell-check tells a reader nothing; a declaration that names the drafting of a Discussion section tells them a great deal.
Every reference, quotation, number, equation and factual claim produced with the assistance of a tool must be checked by the authors against the primary source before submission. A citation is verified when an author has opened the cited work and confirmed that it exists, that the authors and the year are correct, and that it supports the statement it is attached to.
Unverifiable or non-existent citations are treated as a serious integrity failure. Where they are found before publication the manuscript is rejected or returned depending on the extent; where they are found after publication the article is corrected or retracted.
Where a generative model is the subject of the study rather than an aid to writing it, it is described in the Methods, not in the Declaration of Generative AI Use. This is a reproducibility requirement:
See Author Guidelines and Data Availability. Using a model as an object of study does not require a declaration under this policy; using one to write about it does.
Reviewers must not upload a manuscript, any part of a manuscript, or their own report into a generative AI service or any other third-party system. Manuscripts are confidential, and uploading one discloses unpublished work to a party outside the review process, in some cases permanently.
A reviewer may not delegate the assessment to a tool. The value of review lies in the reviewer's own judgement of the method, the evidence and the claims. A reviewer who has used a tool in any capacity in connection with the review must tell the editors what was used and for what, and the editors decide whether the report can be relied on. See Peer Review Process.
The same confidentiality rule binds editors: no manuscript, no reviewer report and no correspondence goes into a third-party generative AI service. Editorial decisions are taken by people, and no decision — desk rejection included — may be produced by a tool. Tools may be used for administrative support only, and only on material that is already public.
| Situation | Response |
|---|---|
| Undisclosed use found before publication, minor | The authors are asked to supply a corrected declaration; the manuscript continues |
| Undisclosed use found before publication, substantial | Rejection |
| Prohibited use found before publication | Rejection, and the authors' institution is notified where the conduct warrants it |
| Undisclosed or prohibited use found after publication | A correction, or a retraction where the reliability of the article is affected, under the COPE framework |
Post-publication procedure is set out in Corrections, Retractions and Expressions of Concern; the ethical framework is described in Publication Ethics.
The journal does not use AI-detection tools as the basis of an accusation. Such tools produce false positives, are unreliable on text written by authors whose first language is not English, and cannot be examined or challenged by the person they accuse. Concerns about generative AI use are pursued through evidence — inconsistent references, results that cannot be reproduced from the described method, figures that do not match the data — and through direct dialogue with the authors, who are shown the specific material and given the opportunity to respond before any decision is taken.
Effective date: 2026-09-01 · Version: 1.0 · Questions: journal@qairu.edu.kz