Artificial Intelligence in Scholarly Publishing



Artificial intelligence (AI) tools, including large language models (LLMs), chatbots, image generators, and other AI-assisted technologies, are increasingly used in research and scholarly publishing. Their use should be consistent with the principles of transparency, accountability, confidentiality, and research integrity.

These Guidelines develop the publication ethics principles applied at the University of Warsaw and take into account guidance from the Committee on Publication Ethics (COPE) , the International Committee of Medical Journal Editors (ICMJE) , and the World Association of Medical Editors (WAME) .

Artificial Intelligence in Scholarly Publishing

Individual journals, series, and publishing units may adopt more detailed rules on the use of AI appropriate to the relevant discipline and type of published material.

1. AI Cannot Be an Author

AI tools cannot be listed as authors or co-authors of scholarly publications.

Authorship entails responsibility for the content of the work, approval of its final version, disclosure of conflicts of interest, and accountability for the accuracy and integrity of the publication. AI tools cannot assume these responsibilities.

See: COPE – Authorship and AI tools and  ICMJE – Use of AI by Authors .

2. Disclosure of AI Use

Authors should transparently disclose the use of AI technologies where they have been used in conducting research or preparing a publication.

Depending on the nature of the work, disclosure may include the use of AI for:

  • generating or editing text;
  • generating images or other graphical material;
  • analysing or processing data;
  • generating computer code;
  • generating tables or other publication content;
  • other substantial activities affecting the content, methodology, or results of the work.

The disclosure should identify the tool used and explain the purpose and manner of its use to a level appropriate to the nature of the work. The relevant journal, series, or publishing unit may specify more detailed disclosure requirements.

Where AI has been used as writing assistance, its use should be disclosed in accordance with the rules of the relevant publication, for example in the acknowledgements or in an appropriate AI-use statement.

See: ICMJE – Use of AI by Authors and  WAME – Chatbots, Generative AI, and Scholarly Manuscripts .

3. Authors' Responsibility

Authors remain fully responsible for the content of their publication, including material prepared, modified, or analysed with the assistance of AI.

Authors should verify, in particular:

  • the factual accuracy of the content;
  • the integrity of data, analyses, and results;
  • the accuracy of citations and references;
  • the originality of the material;
  • compliance with copyright and other third-party rights;
  • the absence of fabricated, manipulated, or misleading information.

The use of AI does not relieve authors of their responsibility to independently review and approve all elements of the publication.

4. Citations, Sources, and Copyright

The use of AI does not remove authors' responsibility to acknowledge sources correctly and respect intellectual property rights.

Authors should independently verify all citations, references, and sources suggested by AI tools. AI systems may generate non-existent, inaccurate, or incomplete references.

Authors are also responsible for appropriately acknowledging reused content and obtaining any permissions required for the use of protected material.

5. AI as Part of Research Methodology, Analysis, and Research Materials

Where AI forms part of the research methodology, data analysis, generation or transformation of results, computer code, tables, images, or other research materials, its use should be appropriately described.

The description should be sufficient to allow assessment of the procedure and, where possible, its reproducibility. Depending on the nature of the research, it may be appropriate to identify the tool, version, purpose of use, and other information necessary to understand how the results were obtained.

See: WAME – Chatbots, Generative AI, and Scholarly Manuscripts .

6. Confidentiality and Unpublished Materials

Unpublished manuscripts, reviews, data, and other materials submitted as part of the editorial process should be treated as confidential.

Editors and reviewers should not upload manuscripts, parts of manuscripts, or other confidential materials to external AI systems where the confidentiality of the information cannot be assured, unless the author has explicitly given permission.

This also applies to the use of AI tools to summarise, analyse, translate, or prepare an assessment of unpublished material.

See: ICMJE – Use of Artificial Intelligence in Publishing .

7. Use of AI by Reviewers

Responsibility for a peer review remains with the reviewer. AI cannot replace the reviewer's own expert assessment of the manuscript.

Reviewers should follow the AI policy of the relevant journal or publishing unit. Where the use of an AI tool in peer review is permitted, such use should be appropriately disclosed.

Regardless of the tools used, the reviewer remains responsible for the final content, quality, and integrity of the review and for maintaining confidentiality.

See: ICMJE – Use of Artificial Intelligence in Publishing and  WAME – Recommendations on Generative AI .

8. Use of AI by Editors and Publishing Units

AI may support selected technical and organisational elements of scholarly publishing but cannot replace editorial accountability.

Final decisions concerning acceptance or rejection should remain under human control. Similarly, assessments of suspected misconduct and decisions concerning corrections, expressions of concern, or retractions should not rely solely on automated AI output.

Where AI is used by an editorial office or publishing unit in a way that affects manuscript assessment or published content, the rules governing such use should be transparent.

See: COPE – Artificial intelligence (AI) in decision making and  ICMJE – Editors’ Role in Ensuring Responsible Use of AI .

9. Publication-Specific AI Policies

Journals, series, and publishing units should establish, to an extent appropriate to their activities, rules governing the use of AI by authors, reviewers, and editors.

A publication-specific policy may address, in particular, permitted uses of AI, disclosure requirements, AI-generated graphical material and data, use of AI during peer review, and the protection of confidential information.

10. Unethical Use of AI

The use of AI may constitute a breach of publication ethics where it results, among other things, in deception, concealment of substantial AI use, a breach of confidentiality, plagiarism, copyright infringement, fabrication or falsification of data, or other forms of misconduct.

Concerns regarding unethical use of AI should be reported and handled under the applicable procedures for publication ethics, taking relevant COPE guidance into account.

University of Warsaw and Polish Academic Publishing Guidance

Responsible use of AI should also take into account guidance developed within the Polish academic publishing community and relevant regulations and guidance of the University of Warsaw.

Useful Resources

Contact

General questions concerning publication ethics at the University of Warsaw may be addressed to:

Research Services Office
E-mail: ja.lewczuk@uw.edu.pl