Generative AI Policy

Purpose and scope

We set these rules for AI use at the Journal of Natural Sciences and Mathematics Research (JNSMR): in manuscripts, in peer review, and in editorial work. They bind authors, reviewers, and editors. They start with the December 2026 issue (Vol. 12, No. 2).

AI tools speed up language editing, coding, and literature searches. They invent references, state errors with confidence, and may store what users type into them. Authors answer for the published article, so we ask you to name the tools you used and to check what they produced.

We add disclosure steps to our existing rules. Fabrication, falsification, and plagiarism were misconduct before AI tools existed, and they remain misconduct when a tool produces the text, data, or image.

Covered tools and uses

These rules cover generative AI: software that produces new text, code, images, or other content from a prompt. Chatbots and large language models (for example ChatGPT, Gemini, Claude, and Copilot), AI paraphrasing and rewriting tools, image generators, and AI coding assistants all count.

We separate AI used to prepare a manuscript from AI used as a research method.

AI as a preparation aid. You used a tool while preparing the manuscript, for example to edit or translate text, summarize papers, write code, or make a figure. The Authors and Disclosure sections apply.

AI as a research method. Your study trains, tests, or evaluates a model, such as a neural-network forecaster, a classifier, or a machine-learning scoring function in a computational study. Describe the model like any other method: software and versions, training and test data, parameters, and code or data availability. The method needs no AI declaration. If you used a generative tool to write the code or the surrounding text, declare that use.

Two kinds of tools fall outside the disclosure rule: rule-based spelling and grammar checkers that mark errors for you to fix, and reference managers such as Zotero or EndNote. If a tool rewrites your sentences or produces text, code, or images you did not create yourself, treat it as generative AI, whatever its marketing label.

Authors

Authorship and responsibility

An AI tool cannot be an author. It cannot approve the final version, answer questions about the work, declare conflicts of interest, or sign the copyright transfer agreement. Do not list a tool as an author, co-author, or contributor.

You answer for everything in your manuscript, including the text, data, code, figures, and references that a tool produced. Read and check all output. AI output can be wrong, incomplete, or biased, and it can reproduce other people’s wording without credit.

Study design, analysis choices, interpretation of results, conclusions, and claims of novelty stay with the authors. Do not hand them to a tool.

Permitted uses, with a declaration

You may:

Edit, rewrite, or translate your own text.
Summarize or search the literature, if you read every source you cite.
Write or debug code, if you test it against known results.
Run an analysis, if you check the output against an independent method.

Declare each use as the Disclosure section describes.

Prohibited uses

You must not:

Fabricate, falsify, or fill in data, results, or observations with AI output, or present generated data as measured data.
Create or alter figures, photographs, micrographs, spectra, or graphical abstracts with generative AI. The exception below applies.
Cite a reference you have not opened and read.
Cite a chatbot’s answer as the source of a scientific fact.
Submit wording or ideas from published work without credit, whether you or a tool wrote it. A paraphrasing tool does not remove the duty to credit a source.
Enter personal data about research participants, other people’s confidential material, or unpublished work you have no right to share into an AI tool.
Hide instructions in a manuscript (white text, metadata, comments) to steer an AI tool or a reviewer. We treat this as misconduct.

Figures and graphical abstracts

JNSMR requires a graphical abstract, so every author must follow this rule. Draw figures and graphical abstracts yourself or with conventional software. We treat a chart that your own code plots from your own data as your work. If an AI tool wrote the plotting code, declare it as code assistance.

You may adjust brightness, contrast, or color balance across a whole image if the change hides nothing present in the original. If a generative model is the subject of your study or a documented part of your method, describe it in the Methods with the model, version, settings, and prompts, and keep the unprocessed outputs. The Editor decides whether such a figure can appear.

Data, code, and references

Label synthetic data. If you generated or augmented data with a model, for a simulation or for training, name the generator in the Methods and keep real and synthetic data in separate tables and figures.

Open every source you cite. AI tools produce references that look real and do not exist, or exist and do not say what the tool claims. We may check references by DOI or title, and we treat a fabricated reference as fabrication.

Keep a record of each substantive AI use: tool name, version, date, and prompts. Give the record to the editors if we ask.

Disclosure

Declare every use of generative AI in preparing your manuscript. Add a section titled “Declaration of AI use” before the References in every manuscript. If you used no generative AI, say so there.

In the declaration, name the tool, its version and developer, the dates of use, the task, and the parts of the manuscript it touched. State that you checked and edited the output and take responsibility for it.

If AI contributed to analysis, code, or figures, describe the use in the Methods in addition to the declaration. Add one line about AI use to your cover letter. For results produced with generative AI, attach a supplementary file with the full prompts, the tool and version, and the date and time of each query.

We do not penalize disclosed use that follows these rules.

Status of each use

Use	Status	What you declare
Rule-based spelling and grammar checkers; reference managers	Allowed	Nothing
Generative AI to edit, rewrite, paraphrase, or translate your text	Allowed	Declaration of AI use
Generative AI to draft text from your own notes or data	Allowed; you write the interpretation and conclusions	Declaration of AI use
Generative AI to find or summarize literature	Allowed; you read every source you cite	Declaration of AI use, if its wording enters the manuscript
Generative AI to write or debug code	Allowed; you test the code	Declaration of AI use and Methods
Generative AI to run an analysis that produces reported results	Allowed after an independent check	Declaration, Methods, and the prompt file
Generative AI to create or alter figures, photographs, spectra, or graphical abstracts	Not allowed, except when the model is your subject of study	Methods: model, version, settings, prompts; the Editor decides
AI or machine learning as your research method	Allowed	Methods, like any method; no declaration for the method itself

Reviewers

You review a confidential manuscript in a closed double-blind process, and the report is your own work.

Do not upload a manuscript, any part of it, or your report to a generative AI tool, including to polish your language. Such services may store or reuse what you enter, and the authors have not agreed to that. This covers AI detectors.
Do not use generative AI to read, summarize, assess, or draft your review. You make the scientific judgment and answer for what the report says.
If you think a tool would help in a specific way, ask the handling editor first and wait for written permission. With permission, state the use in your report and check the output for errors and bias.
If you suspect undisclosed AI use, tell the editor in your confidential comments and give the evidence, such as references you could not find.

Editors and the journal

Editors make the decisions on manuscripts. No AI tool makes, recommends, or drafts a decision.

Editors and editorial staff must not upload manuscripts, review reports, decision letters, or confidential correspondence to a public generative AI tool, including to edit language. They tell the Editor-in-Chief about any AI use in handling a manuscript.

Our Peer Review page states that we check submissions with similarity-checking software. We add another AI-based tool only if the vendor’s terms protect confidentiality and bar training on our content, and a human checks each flag before anyone acts on it. We list each such tool on this page.

We do not treat an AI-detection score as evidence. A 2023 study in Patterns found that popular detectors misclassified writing by non-native English speakers as AI-generated, and many JNSMR authors write in English as a second language.

We act on verifiable evidence, such as:

references that do not exist
figures with signs of synthesis
text that contradicts the reported data
an author who cannot produce sources or explain the method

Violations

We handle suspected undisclosed or prohibited AI use through the process on our Allegations of Research Misconduct page. That process uses COPE guidance.

The editors check the concern against evidence. We do not accept a detector score as evidence.
We write to the corresponding author with the concern and ask for an explanation and supporting material, such as prompts, drafts, source data, and code, by a date we give.
The Editor-in-Chief decides after consulting an editorial board member who has no conflict of interest.

The outcome depends on the stage and the severity:

Omitted declaration, findings unaffected: we ask for the declaration and, for a published article, issue a correction notice.
Prohibited use or fabricated content in a manuscript under review: we reject the manuscript.
Fabricated data, figures, or references in a published article: we retract it and follow the COPE retraction guidelines.
Suspected misconduct: we notify the authors’ institutions. For severe or repeated cases, the Editor-in-Chief may decline future submissions from the authors.

You may appeal in writing to the Editor-in-Chief and supply evidence. A board member who did not handle the case reviews the appeal.

Effective date and transition

These rules apply to every article we publish, starting with the December 2026 issue (Vol. 12, No. 2).

New submissions: include the declaration when you submit.
Manuscripts under review on the day we publish this policy: include the declaration with your next revision.
Manuscripts we accepted before that day: send the declaration before we release the proofs. If a figure or graphical abstract falls under the prohibited uses, we ask you to replace it before publication.

The rules against fabrication, falsification, and plagiarism apply to every manuscript, whatever its date.

Basis, review, and contact

We based these rules on three sources:

COPE, Authorship and AI tools
ICMJE Recommendations, Section V: Use of AI by authors, by reviewers, and editors’ role
WAME, Chatbots, Generative AI, and Scholarly Manuscripts, revised May 31, 2023

We drew the rule on detection scores from Liang W, Yuksekgonul M, Mao Y, Wu E, Zou J. GPT detectors are biased against non-native English writers. Patterns, 2023.

We review these rules once a year and after any change to COPE or ICMJE guidance. We give each revision a version number and a date at the foot of this page. If you are unsure whether a use complies, write to the editorial office through the Contact page before you submit.

Version 1.0. The editorial board approved it on October 2026. We apply it from Vol. 12, No. 2 (December 2026).

Appendix: AI use declaration

Paste one of these forms under the heading “Declaration of AI use,” before the References.

If you used generative AI:

Declaration of AI use. [Author initials] used [tool name, version, developer] between [start date] and [end date] to [task, for example: translate the Introduction from Indonesian into English]. The affected parts are [sections]. [Author initials] read, checked, and edited all output and take responsibility for the final text, data, code, and references.

If you used none:

Declaration of AI use. The authors did not use generative AI tools in preparing this work.

For results produced with generative AI, attach a supplementary file that lists the full prompts, the tool and version, and the date and time of each query.