This policy of Medicine, Science and Education sets out the conditions for using generative artificial intelligence in manuscript preparation, editorial assessment, and peer review. The use of such tools does not relieve authors, editors, or reviewers of personal responsibility for the content, scientific assessment, or decisions.
1. Disclosure by Authors
Authors must disclose the use of generative artificial intelligence in both the cover letter and the manuscript, specifying:
- the tool name and available version;
- the purpose and stage of use;
- the human verification performed.
AI assistance with language preparation must be described in a declaration. Its use for data analysis, programming, image-related work, or other research procedures must be reported in the Methods section in sufficient detail to allow verification.
If generative AI was not used, this must be stated in the relevant declaration. Basic spelling checks that do not generate content do not require separate disclosure.
2. Authorship and Accountability
Artificial intelligence must not be included in the author list.
Authors must personally verify facts, references, calculations, code, translations, and compliance with third-party rights. They remain responsible for the accuracy and integrity of the submitted materials, regardless of the tools used.
3. Data and Images
Generating or altering research data or images in a way that misrepresents facts is prohibited.
Where synthetic data or generative methods are themselves the subject of research, their nature, methods of creation, and limitations must be explicitly described.
4. Use by Editors and Reviewers
Editors and reviewers must not upload unpublished manuscripts, personal information, or confidential review reports to publicly accessible AI services.
An approved secure tool may be used only after verifying:
- confidentiality safeguards;
- the right to process the materials;
- that the materials will not be used to train the system;
- completion of any necessary agreement with the authors.
The use of such a tool must be documented. Scientific assessment and editorial decisions remain a human responsibility.
5. Assessment of Potential Violations
An automated AI-detector score does not, by itself, constitute evidence of a violation.
Where concerns arise, the editorial team requests explanations, underlying data, and the history of manuscript preparation. Decisions are based on the totality of verifiable circumstances. Substantiated concerns are handled in accordance with the journal’s procedures.