Generative AI policies for journals

For authors

Using generative AI and AI-assisted technologies when preparing a manuscript: an overview

JTKL accepts that AI tools can help researchers work more efficiently, reach insights sooner and achieve better results, provided they are used responsibly.

Such tools, including AI agents and deep research tools, are increasingly used to digest complex literature, outline a field or research question, spot gaps in knowledge, generate ideas, and assist with organizing content and improving language and readability. Authors preparing a manuscript for a JTKL journal may use AI tools for these purposes. They must not, however, take the place of the author's own critical thinking, expertise and judgement.

AI tools should always be used under human oversight and control. In the end, authors remain responsible and accountable for what their work contains. This means:

Carefully checking the accuracy, completeness and impartiality of everything an AI tool produces, including verifying the sources, since AI-generated references may be wrong or invented.

Thoroughly editing and adapting all material so the manuscript is the author's own original contribution and expresses their own analysis, interpretation, insight and ideas.

Being open about any tools or sources used, AI-based or not. For AI tools, a disclosure statement is required at submission.

Developing the manuscript in a way that protects data privacy, intellectual property and other rights, by reading the terms and conditions of every AI tool used.

Responsible use of AI tools
Before using an AI tool, authors must read its terms and conditions to be sure that the privacy and confidentiality of their data and inputs, unpublished manuscripts included, will be protected. Special care is needed with personally identifiable data. Authors should also look for factual errors and possible bias in the output.

Authors should further confirm that the terms give the tool only the right to use their material to deliver the service to them, and no other rights over what they enter (in particular, no right to train the AI model on it). They should also confirm that the tool places no limits on how its output may be used that could hinder later publication of the article.

Disclosure
When AI tools were used to prepare the manuscript, authors should say so in a separate AI declaration statement included with the submission. The statement will be printed in the published article for transparency. Authors should keep a record of their AI use, noting the name of the tool, why it was used and how closely they supervised it. Declaring AI use builds transparency and trust among authors, readers, reviewers, editors and contributors, and helps meet the terms of use of the tool concerned.

Note: routine checks of grammar, spelling and punctuation need no declaration. If an AI tool makes substantial changes to the sentence structure or organization of any part of the text, however, this should be disclosed.

Authors are also not obliged to disclose AI features built into specialist assistive technology for people with disabilities, as long as they are used only for accessibility.

If AI tools were part of the research itself rather than the writing of the manuscript, that use should be described in detail in the Methods section.

The recommended wording for the declaration is:

Section title: Declaration of generative AI and AI-assisted technologies in the manuscript preparation process

Statement: While preparing this work, the author(s) used [NAME OF TOOL / SERVICE] to [REASON]. Afterwards, the author(s) reviewed and edited the content as necessary and accept(s) full responsibility for the content of the published article.

Authorship: AI tools must not be named as an author or co-author, and must not be cited as an author. Authorship carries duties and tasks that only humans can hold. Every (co-)author must make sure that any question about the accuracy or integrity of any part of the work is properly looked into and resolved, and authorship requires being able to approve the final version and agree to its submission. Authors must also ensure that the work is original and not published before, that the named authors qualify for authorship, and that the work does not infringe the rights of third parties. They should read JTKL's Ethics in Publishing policy before submitting.

Using generative AI and AI-assisted tools for figures, images and artwork
AI tools may be used to create figures, images and artwork only in certain situations.

Images fall broadly into three categories, and what is allowed differs for each.

(1) Explanatory images
Explanatory images illustrate concepts, processes or relationships, such as flow charts, decision trees, timelines or schematic conceptual drawings.

Authors may use AI tools to help produce some kinds of explanatory image for publication, for example flow charts, decision trees, timelines, schematic concept illustrations or experimental workflow diagrams.

When AI tools are used this way, the principles of responsible use, human oversight, transparency, disclosure and authorship described under "Authorship" apply.

The use of AI must be disclosed in the caption of each such image (naming the tool, its version and how it was used) and in the article's general AI disclosure statement.

(2) Research and data images
Research and data images present research data or results. They comprise (a) data visualizations, produced from underlying datasets through analytical, computational or statistical workflows (plots, charts, graphs, heatmaps and the like), and (b) primary research images, which show primary observed or experimental data (microscopy, histology, western blots, radiology scans, patient images and the like).

2a) Data visualizations
Data visualizations show results derived from underlying datasets through analytical, computational or statistical workflows.

AI tools must never be used to fabricate results, invent or change underlying data, or produce figures that are not faithfully derived from the data and methods used in the research.

AI tools may help create data visualizations only if the output comes directly from the underlying data. Such visualizations must be produced by reproducible analytical, computational or statistical methods that are clearly reported in the Methods section. If AI tools are involved, authors should state the name and version of the model or tool and its developer or manufacturer in the Methods section, follow the tool's usage policies and give proper attribution where required. Authors may be asked to supply documentation of their AI use and the original, unprocessed images for editorial review.

2b) Primary research images
Primary research images show primary observed or experimental data (microscopy, histology, western blots, radiology scans, patient images and the like).

AI tools must not be used to create or modify images that represent primary observed or experimental data not directly obtained in the research. This includes changes to brightness, contrast or color balance, which should be made only with established image-processing software. See the artwork instructions for more detail.

Use of AI tools in research methods

This policy does not stop AI tools from being used in formal research design or methods, for instance AI-assisted imaging used to collect or interpret research data, or predictive modelling of clinical outcomes from patient data. Where AI is part of the research design or methods, its use must be described reproducibly in the Methods section, including the name and version of the model or tool and, where relevant, its developer or manufacturer. Authors should follow the tools' usage policies and attribute them properly where required.

(3) Graphical abstracts and cover art
Graphical abstracts and cover art are made to summarize or represent an article visually and are not part of its scientific content.

General-purpose generative AI image tools must not be used to create graphical abstracts, which are often complex summaries combining several graphical elements, icons and design components.

Authors are encouraged to use dedicated scientific illustration tools for graphical abstracts, as these have clear licensing terms that support responsible handling of copyright and intellectual property.

AI-generated journal cover art may sometimes be allowed. Authors must get permission from the journal editor and publisher beforehand, show that all necessary permissions for third-party material have been obtained, and provide proper attribution.

Author responsibilities
Authors must ensure that every image they submit is accurate and original. That includes checking that the concepts shown are correct and that the images reflect the author's own work and ideas. Where an image builds on existing artwork or graphics, it must be properly credited and permission obtained from the rights holder before submission. Images that copy or refer to existing copyrighted images, real people, or other parties' identifiable products or brands must not be generated, nor may any likeness of a person's voice.

For reviewers
Using generative AI and AI-assisted technologies in journal peer review
Peer review is central to the scientific system, and JTKL maintains the highest standards of integrity in it. A manuscript a researcher is asked to review is a confidential document and must be handled as such.

Reviewers should not upload a submitted manuscript, or any part of it, to an AI tool, since that could breach the authors' confidentiality and proprietary rights and, if the paper contains personal data, data privacy rights.

Reviewers are responsible for the scientific assessment and judgement of the manuscripts they evaluate. AI tools cannot replace a reviewer's critical thinking or independent evaluation, because the technology can produce incorrect, incomplete or biased conclusions. Reviewers remain fully accountable for their reports.

AI tools may therefore be used only in a supporting role, for example to improve the language and structure of a report or to help with a background literature search, as long as confidentiality is preserved and humans stay in control.

Disclosure
Reviewers should state their AI use in their review reports, naming the tool and the purpose. Basic spelling and grammar checks need not be disclosed.

Suggested statement:

While preparing this report, I used [NAME OF TOOL / SERVICE] to [REASON]. Afterwards I reviewed and edited the content as necessary and take full responsibility for it.

What is a private AI tool?
A "private" AI tool is a version that does not keep, reuse, share or learn from the content you give it. Anything you paste or upload should not be retained beyond what is needed to provide the service, and must not be used to train or improve the model or be shared with third parties.

Before using any AI tool, check the following:
Read the privacy policy and terms of service, paying particular attention to data retention and model training. A private tool will say clearly that inputs are not kept beyond what the service needs and are not used for training. Users should also make sure the tool grants the rights needed for publication.

Favor paid or enterprise versions over free ones. Many free AI tools are not private by default. Check whether your institution offers an enterprise version or a locally hosted model. These are more likely to include privacy protection, though this is not guaranteed and should still be verified.

For editors
Using generative AI and AI-assisted technologies in the editorial process
Academic editors are vital in upholding the journal's profile and reputation. Overseeing the editorial evaluation of a scientific manuscript involves responsibilities that only humans can carry.

While a manuscript is being evaluated, it must be treated as confidential.

Editors should not upload a submitted manuscript, or any part of it, to an AI tool, since that could breach the authors' confidentiality and proprietary rights and, if the paper contains personal data, data privacy rights.

Editors are responsible for the editorial assessment and the decision process. AI tools cannot replace an editor's critical thinking, independent evaluation or final decision. Editors stay fully accountable for the process, the decision and how it is communicated to authors.

Editors may therefore use AI tools only in a supporting role, for example to improve the language or structure of decision letters or to help with a background literature search, provided confidentiality is kept and humans remain in control.

How JTKL currently uses AI in publishing

Our teams may use AI tools to help with:

Technical checks of submitted papers, such as compliance with submission guidelines and completeness.

Research integrity checks to confirm that our policies are followed.

The stage after acceptance, including proof preparation, copy editing, and finding inconsistencies or errors in the final paper. We are committed to keeping human oversight at the center of decision-making.