Jurnal Civics: Media Kajian Kewarganegaraan recognises that generative artificial intelligence (AI) tools can support scholarly work, for example in improving language, supporting non-native English writers, generating ideas, and speeding up research and dissemination. Large language models and multimodal models such as ChatGPT, Copilot, Gemini, Claude, DALL-E, and Midjourney are examples. These tools also carry risks. This policy sets out how authors, reviewers, and editors may use generative AI responsibly. It is based on the COPE position statement on authorship and AI tools and on current international publishing practice, and will be revised as the technology and good practice develop.

1. Why Caution Is Needed

When using generative AI, users should be aware of four main risks:

• Inaccuracy and bias: AI tools are statistical, not factual. They can produce false information (“hallucinations”), fabricated references, and biased content. Bias deserves particular attention in citizenship and civic education research, which often concerns gender, religion, ethnicity, minorities, and marginalised groups.
• Lack of attribution: AI output often reproduces others’ ideas or text without proper credit, which may amount to plagiarism.
• Confidentiality and intellectual property: content entered into third-party AI platforms may be exposed or misused.
• Unintended reuse: AI providers may reuse input or output data, for example to train their models.

2. Policy for Authors

2.1 Accountability

• Authors are fully responsible for the originality, validity, and integrity of their manuscript, including any part produced with AI assistance.
• Authors must review all AI output carefully, check facts and references against the original sources, and make sure that the work reflects their own analysis and argument.
• Before using a tool, authors should check that it is suitable for the intended purpose and that its terms of use adequately protect intellectual property, confidentiality, and security. This matters especially for interview transcripts and other data from human participants.

2.2 Authorship

• Generative AI tools must not be listed as an author or co-author, and must not be cited as an author. AI cannot take responsibility for the content, manage copyright and licensing, consent to publication, or give contractual assurances. These are all requirements of authorship.

2.3 Permitted uses (with disclosure)

• Idea generation and exploration, such as brainstorming research questions or outlining.
• Language improvement, copyediting, and proofreading, including support for non-native English writers.
• Translation of the author’s own text, for example from Indonesian into English, provided that the author checks the result carefully.
• Literature search and classification, provided that every source is checked against the original publication.
• Coding assistance, for example in R, Python, or NVivo/ATLAS.ti scripts, provided that the code is checked and validated.

2.4 Prohibited or high-risk uses

• Generating text or code without thorough human revision, or submitting AI-generated text as the authors’ own analysis or argument.
• Generating or altering research data, results, interview transcripts, quotations, survey responses, or field notes, or creating synthetic data to replace missing data without a robust, transparent, and disclosed methodology.
• Generating references or citations. Every reference must be a real, verified source that the authors have read.
• Producing inaccurate content, including abstracts, keywords, or supplementary materials.
• Uploading confidential or identifiable participant data to AI tools without the consent and ethics approval that cover this use.

Submissions involving prohibited uses may be investigated under the journal’s Research Misconduct policy, and may be rejected or, if already published, corrected or retracted.

2.5 Images and figures

• Permitted: AI assistance in creating data visualisations, conceptual illustrations, flow diagrams (e.g., a PRISMA flow diagram), and explanatory figures, provided that this is disclosed and the underlying data are accurate.
• Not permitted: using AI to create or manipulate research outputs, such as photographs from fieldwork, documentary evidence, screenshots, or other images presented as data. Manipulation includes enhancing, obscuring, moving, removing, or adding features in an image.

2.6 Disclosure requirements

Authors must include a Declaration of Generative AI Use in the end-matter of the manuscript, as provided in the journal templates. The declaration must state:

1. the name and version of the tool (e.g., ChatGPT-4o, Claude, Gemini, DeepL Write);
2. how it was used, naming the sections affected;
3. why it was used;
4. how the output was reviewed and verified by the authors.

If no generative AI was used, authors state: “The authors did not use generative AI tools in the preparation of this manuscript.” Basic spelling and grammar checkers, and reference managers such as Mendeley or Zotero, do not need to be declared.

3. Policy for Reviewers

• Reviewers must not upload manuscripts under review, or any part of them, to generative AI tools. This includes files, figures, tables, and supplementary information. Doing so breaches the confidentiality of the authors’ work and the anonymity of double-anonymous review.
• Reviewers must not use AI tools to analyse or summarise submitted manuscripts, or to generate review reports.
• Reviewers may use AI tools only to improve the language of a review they have written themselves, without entering any content from the manuscript.
• Reviewers remain fully responsible for the accuracy and integrity of their reviews.
• If reviewers suspect undisclosed or inappropriate AI use in a manuscript (e.g., fabricated references or generic, unsupported text), they should tell the editor in their confidential comments rather than making accusations in the review.

4. Policy for Editors

• Editors must keep submissions and peer-review information confidential. They must not upload unpublished manuscripts, associated files, images, reviews, or decision letters to generative AI tools.
• Editors must not use AI to make or justify editorial decisions. Decisions are made by human editors.
• Editors may use only tools that the journal has approved for specific tasks, such as similarity checking, reference validation, or detection of image manipulation. These tools must comply with confidentiality and data-protection requirements.
• Results from AI-detection tools are not proof of AI use. Editors treat them as one indicator among others and give authors the opportunity to explain before taking any action.

5. Handling Breaches

1. Identification. A concern is raised by an editor or reviewer, by screening (e.g., non-existent references, tortured phrases, inconsistent content), or by a reader.
2. Request for explanation. The editor asks the authors to explain their use of AI and to provide evidence, such as drafts, data, or the sources behind the references, within 14 days for manuscripts under review or 30 days for published articles.
3. Assessment and action.
   • Undisclosed but acceptable use: the authors must add a declaration (before publication) or a corrigendum is published (after publication).
   • Prohibited use: the manuscript is rejected, or the published article is retracted. The case may be handled under the Research Misconduct policy, including notifying the authors’ institutions.
   • Breaches by reviewers or editors: the review is disregarded, and the person is removed from the reviewer database or editorial board.

Sources

• Committee on Publication Ethics. (2023). Authorship and AI tools: COPE position statement. https://publicationethics.org/cope-position-statements/ai-author
• International Committee of Medical Journal Editors. (2024). Recommendations for the conduct, reporting, editing, and publication of scholarly work in medical journals (Section II.A.4: Artificial intelligence (AI)-assisted technology). https://www.icmje.org/recommendations/
• World Association of Medical Editors. (2023). Chatbots, generative AI, and scholarly manuscripts: WAME recommendations on chatbots and generative artificial intelligence in relation to scholarly publications. https://wame.org/page3.php?id=106