General Principles
-
Humans remain the party responsible for the final content of scientific documents.
-
The use of AI must be transparent if it provides a substantive contribution to the preparation of a manuscript or review report.
-
All AI outputs must be checked, verified, and edited by humans before use.
-
Confidentiality, privacy, intellectual property rights, and publication rights must be protected.
-
AI must not be used to falsify, fabricate, or alter research evidence that was not genuinely obtained.
-
AI must not replace human scientific judgment, whether by authors or reviewers.
Prohibited Uses of AI for Authors
1. Delegating scientific responsibility to AI
AI must not make final scientific decisions, determine the validity of hypotheses without human evaluation, or generate conclusions that are accepted without review.
2. Creating an entire manuscript with AI and claiming it as the author's work
Scientific work must reflect the authors' original analysis, interpretation, and contribution. AI outputs must not replace human intellectual contribution.
3. Using fake or unverified references
It is prohibited to use DOIs, titles, author names, journals, or publications that are solely provided by AI without verification against the original source.
4. Falsifying data
AI must not be used to create respondents, add samples, fabricate interviews, generate experimental values, or fill in data that were never actually obtained.
5. Manipulating results
AI must not be used to alter data or analysis in order to achieve desired results, conceal contradictory findings, or force interpretations not supported by the data.
6. Creating or altering primary data images
For images that constitute research evidence—such as microscopy, histology, western blots, radiology scans, patient images, or experimental imagery—AI must not be used to create or alter information that was not actually observed.
7. Creating graphical abstracts with general-purpose generative AI when prohibited by the journal
The latest Elsevier policy states that general-purpose generative AI image tools must not be used to create graphical abstracts. The specific provisions of the target journal must still be checked.
8. Uploading confidential data to unsecured AI systems
Personal data, patient data, identifying information, confidential company data, manuscripts of other parties, or materials bound by confidentiality must not be entered into AI systems without adequate protection of privacy, IP, and publication rights.
9. Disguising plagiarism
AI must not be used to paraphrase others' work in order to pass similarity checks or avoid attribution obligations.
10. Concealing substantive use of AI
If AI provides a meaningful contribution to the preparation of a manuscript, its use must be disclosed in accordance with journal policy.










