The use of generative artificial intelligence in preparing publications is no longer unusual, and major publishers – Elsevier, Springer Nature, Wiley – have formalised the rules for its use in official policies. Since 2023, disclosure requirements for AI use have gradually tightened, and by 2026 a stable set of standards has taken shape that every author should know before submitting an article. The situation in Poland deserves particular attention: the national system for evaluating scientific activity is directly linked to the list of journals maintained by the Ministry of Science and Higher Education, and checking publications for compliance with ethical standards is becoming part of the new evaluation methodology for 2026-2030.
Why did journals introduce AI declarations?
Widespread use of tools such as ChatGPT, Gemini or Claude in preparing manuscripts has left editorial boards with a practical problem: how to distinguish an author's own work from AI-generated content, and who bears responsibility for the content of an article. Almost all major publishers have arrived at the same answer – artificial intelligence cannot be listed as an author of a publication, since it cannot take responsibility for the accuracy and originality of the material. This principle is set out by Elsevier, Springer Nature, and in the guidelines issued by the ICMJE and COPE.
At the same time, the use of neural networks during manuscript preparation is not generally prohibited. This covers language editing, structuring the text, working with code, or an initial literature review. The main condition: any such use must be disclosed, and the author must take responsibility for the final text.
What should a declaration of AI use contain?
The format of the declaration varies between journals, but the set of required elements is largely consistent across most publications. The section is usually placed before the reference list, under a heading such as “Declaration of generative AI use”, and should specify:
- the name of the tool and its version
- the specific task for which the tool was used
- the section of the manuscript where this use is reflected
- confirmation that the author reviewed the output and takes responsibility for it
A vague statement such as “the text was edited using AI” no longer satisfies most editorial boards. Authors are expected to provide specifics: which tool, for what purpose, and what exactly was checked manually.
What restrictions apply to images and data?
A separate category of rules concerns visual material. Most journals, including The BMJ and Elsevier titles, explicitly prohibit the use of AI-generated images, illustrations and graphical material in scientific articles. This requirement also extends to figures produced using neural networks to present research results. Authors must also ensure that the tool used grants the rights required for publishing the final material.
Publication policy on data remains unchanged: any processing of results using AI must be described in the methods section, alongside other analytical tools.
How are journal AI policies classified?
Despite the shared principle that AI cannot be an author, publishers' approaches to other questions differ considerably. They can be divided into three groups:
- Journals that permit AI use provided it is fully disclosed and the author retains responsibility for the final content – this is the dominant approach in 2026, followed by Elsevier, Springer Nature, Wiley, Taylor & Francis and SAGE.
- Journals that limit AI use to specific tasks – for example, language editing only – while still requiring a declaration even in this case.
- Journals and funding bodies that treat AI-generated text as incompatible with the principles of originality and authorship, rejecting such manuscripts regardless of disclosure.
Before submitting an article, it is worth establishing which of these groups the chosen journal falls into, as this determines which wording is acceptable in the text and where the declaration should be placed.
How do the new requirements affect authors in Poland?
For authors publishing in Poland, publication ethics is tied not only to the policy of a given journal but also to the national system for evaluating scientific activity. The Rada Doskonałości Naukowej (National Council for Scholarly Development) oversees the procedures for awarding doctoral and habilitation degrees, and publications in journals from the official list of the Ministry of Science and Higher Education are taken into account when evaluating research output.
In 2026, the ministry is completing an updated list of journals for the 2026-2030 evaluation period, and for the first time the new methodology includes a mechanism for reducing points or removing a journal from the list where ethical or scientific standards have been breached.
This means that, for authors in Poland, choosing a journal is a question of its formal compliance with the national list. Publishing in a journal with questionable review practices risks the points not being credited at the next evaluation, even if the journal is indexed in international databases.
Compliance with AI declarations and publication ethics standards in 2026 is a condition for passing peer review and for having a publication counted in scientific evaluation. Authors should study the policy of their chosen journal in advance, describe their use of tools precisely, and confirm that the journal is included in the current national list.
Do you need support preparing a publication that meets current requirements for AI declarations and publication ethics? Specialists at Scientific Publications will advise on journal selection and manuscript preparation for peer review. To receive a free consultation, complete the form below, and our manager will contact you shortly.
