 |
GENERATIVE AI POLICIES
This policy explains how authors, editors, and peer reviewers of Jurnal Ilmu Komunikasi may use generative AI tools, and how they may not use them. It draws on the recommendations of STM, Elsevier, and the World Association of Medical Editors (WAME), and it will be revised when the technology and the standards change.
The short rule is this: humans write, verify, and take responsibility; AI may assist, and its use must be disclosed.
Contents
| No. | Section |
|---|---|
| 1 | Introduction |
| 2 | Policy at a Glance |
| 3 | Authors |
| 4 | Editors and Peer Reviewers |
| 5 | Handling of Violations |
| 6 | Related Pages and Enquiries |
1. Introduction
MUKASI recognizes that generative AI can help with the development of ideas, with the workflow of research, with the analysis of data, with the quality of writing, and with the work of authors whose first language is not English. The journal also recognizes that these tools can cause harm when they are used without care.
Definition. Generative AI means tools that produce text, images, audio, or synthetic data. Examples are ChatGPT, Copilot, Gemini, Claude, NovelAI, and Jasper AI.
Main risks
| No. | Risk | Explanation |
|---|---|---|
| 1 | Inaccuracy and bias | A tool can produce content that is wrong, biased, or invented, which is often called a hallucination. |
| 2 | Weak attribution | Ideas, quotations, and citations may be given without a correct source. |
| 3 | Confidentiality and intellectual property | Material that is entered into a platform of a third party can be exposed. |
| 4 | Unintended secondary use | The provider of a tool may reuse the inputs and the outputs to train its models. |
2. Policy at a Glance
| Use of generative AI | Status | Condition |
|---|---|---|
| Improving grammar, language quality, and readability, and translating the text of the authors | Allowed | The authors verify the output and disclose the use. |
| Generating ideas, searching and categorizing literature, and supporting coding | Allowed | The authors verify the output and disclose the use. |
| Studying AI tools, or using them in a study, an experiment, or an instrument | Allowed | The use is described in the Method section. |
| Listing an AI tool as an author | Not allowed | None. |
| Submitting AI text or code that the authors have not reviewed and edited | Not allowed | None. |
| Replacing missing empirical data with synthetic data without a sound method | Not allowed | None. |
| Creating or altering images, figures, or original research data | Not allowed | The only exception is described in section 3.5. |
| Citing references generated by AI that have not been verified | Not allowed | None. |
| Uploading unpublished manuscripts or related files to AI tools (editors and reviewers) | Not allowed | None. |
| Using AI to evaluate, summarize, or assess a manuscript (reviewers) | Not allowed | AI may be used only to improve the language of the review report. |
3. Authors
3.1 Permitted use
- Authors may use tools based on GPT, such as ChatGPT, to improve grammar, language quality, and readability.
- Authors may use generative AI to develop ideas, to improve language and writing, to search the literature interactively with a large language model, to categorize the literature, and to support coding.
- Authors may translate their own text, for example the abstract, but they must check that the translation is accurate.
3.2 Accountability and human authorship
- Authors remain fully responsible for the originality, the accuracy, the validity, and the ethics of the work.
- Authors must verify all content that was produced with the help of AI, and they must follow the policies on Authorship and Contributorship and Publication Ethics.
- The work must be written by humans in all its essential parts, and it must meet scholarly standards. Human oversight and transparency are expected in the whole research process.
- MUKASI makes the final decision on publication.
3.3 AI tools cannot be authors
An AI tool must not be listed as an author. A tool cannot take responsibility for a work, cannot consent to publication, and cannot sign a copyright agreement or a license agreement.
3.4 Prohibited reliance on AI
A manuscript must not contain any of the items below. Such practices can lead to an editorial review or to an investigation.
| No. | Prohibited content |
|---|---|
| 1 | Text or code produced by AI that the authors have not reviewed and edited. |
| 2 | Synthetic data in place of missing empirical data, unless a sound method is used and described clearly. |
| 3 | Inaccurate content produced by AI, including in the abstract and in the supplementary material. |
3.5 Images, figures, and data
- Generative AI must not be used to create or to alter images, figures, or original research data that are meant for publication. This covers charts, tables, parts of images, code, and formulas.
- To add, remove, hide, or move an element of an image or a figure is an alteration.
- Exception for communication research. Material generated by AI may appear in an article when it is itself the subject or the stimulus of the study, for example an AI generated image in an experiment on media effects, or a chatbot response that is analyzed. Such material must be labeled clearly as AI generated, and the tool, its version, the prompt, and the date of use must be described in the Method section.
3.6 References and citations
- AI can produce references that look real but do not exist.
- Authors must check every reference and every DOI against the original source.
- Fabricated or falsified references are treated as research misconduct, in accordance with the Plagiarism and Research Integrity Policy.
3.7 Research about AI in communication
A study of AI is different from the use of AI to prepare a manuscript. Examples are research on chatbots, on AI in journalism, public relations, or advertising, on deepfakes, on algorithmic news, and on the use of AI tools by audiences or by media workers. The same applies when AI is part of the method, for example to code content, to transcribe interviews, or to create stimulus material for an experiment.
In these cases, the authors must describe the following in the Method section, so that other researchers can evaluate the study.
| Item | What to report |
|---|---|
| Tool and version | The name of the tool, the version or model, and the date of use. |
| How it was used | The task, the prompts or instructions, and the relevant settings. |
| Validation | How the output was checked by humans, for example by a comparison with coders who are human, with a measure of agreement. |
| Limitations | Known errors, bias, or limits of the tool that may affect the findings. |
3.8 Confidentiality and participant data
- Authors must not upload personal or identifiable data of respondents, informants, or other participants, or unpublished research data, to a generative AI tool. This includes interview recordings and transcripts that are sent to an online service for transcription or analysis.
- The exception is the case where the participants have agreed to it and the terms of the tool give adequate protection.
- Authors must confirm that the terms of any tool protect intellectual property, confidentiality, and the security of data.
3.9 Disclosure
Every use of generative AI in the preparation of a manuscript must be disclosed in the manuscript. The statement gives the name and version of the tool, how it was used, and the purpose.
| If the tool was used for | Place the statement in |
|---|---|
| The research process, for example coding support or the categorization of literature | The Method section |
| Language editing or translation | The Acknowledgments section |
Example of a disclosure statement
During the preparation of this manuscript, the authors used [name and version of the tool] to [purpose of use]. The authors reviewed and edited the output and take full responsibility for the content of the manuscript.
Where the statement belongs in the manuscript is also shown in the Author Guidelines and in Manuscript Preparation.
4. Editors and Peer Reviewers
Unpublished manuscripts, images, files, and related information must not be uploaded to generative AI tools. Doing so may break the confidentiality of the review and the intellectual property rights of the authors.
| Role | Rules |
|---|---|
| 4.1 Editors | Editors protect the confidentiality of submissions and of the review process. They do not upload unpublished manuscripts or related materials to generative AI platforms. |
| 4.2 Peer reviewers |
|
See also the Peer Review Process.
5. Handling of Violations
- A suspected violation of this policy is examined in accordance with the Allegations of Misconduct policy.
- The result depends on the seriousness of the case. It may be a request for clarification or for revision, or the rejection of the manuscript. For a published article, the journal acts under the Correction and Retraction policy.
- A conflict of interest that is related to AI, for example an interest in the company that owns a tool used in the study, must be declared under the Conflicts of Interest policy.
6. Related Pages and Enquiries
| Questions about this policy | jurnal.mukasi@gmail.com | jurnal@yp3a.org |
| Preparing a manuscript | Author Guidelines and Manuscript Preparation |
| Ethics and integrity | Publication Ethics, Plagiarism Policy, Authorship and Contributorship |
| Review of manuscripts | Peer Review Process |
| Misconduct, interests, and errors | Allegations of Misconduct, Conflicts of Interest, Correction and Retraction |
