Ethics and Integrity for AI Tools and Emerging Modern Technologies
1. Purpose
We use AI and modern technology to make publishing faster, fairer and more reliable. We never use them to replace human judgment. Our ethical standards follow COPE, ICMJE, WAME, STM, the UNESCO Recommendation on the Ethics of Artificial Intelligence and the OECD AI Principles.
2. Ethical principles
Human accountability: only people can be authors, reviewers or decision-makers.
Transparency: any use of AI that shapes content, analysis or decisions is disclosed.
Confidentiality: unpublished work and personal data are never entered into public AI tools.
Accuracy: AI output is treated as unverified until a qualified person has checked it.
Fairness: AI must not disadvantage anyone because of language, region, institution, career stage or identity.
Proportionality: the level of control matches the level of risk.
Privacy and security by design: data protection is built into the selection and use of every tool.
3. Authorship and accountability
AI tools cannot be authors. They cannot approve a manuscript, take responsibility for it, declare conflicts of interest or hold copyright. All authors are fully responsible for the whole manuscript, including any content produced with AI.
4. Obligations of authors
Check every claim, quotation, data point and reference produced or suggested by AI against primary sources.
Do not use AI to fabricate, falsify or manipulate data, images or results.
Do not use hidden text or prompts to influence AI screening or AI-assisted review.
Make sure AI-assisted content does not reproduce others' work without proper attribution.
Process research participants' data with AI only with research ethics board approval.
5. Obligations of peer reviewers
Peer review must reflect the reviewer's own expert judgment. Reviewers keep manuscripts confidential and never enter them into public AI tools. They do not use generative AI to write their evaluation or recommendation.
6. Obligations of editors
A named human editor makes every editorial decision. Editors use only approved tools. They check the identity, expertise and conflicts of interest of every reviewer suggested by software. They never base a decision on an AI-detector score alone.
7. Fairness and bias control
Authors who write in English as an additional language may use AI for language editing and translation. Disclosing this does not affect the editorial assessment.
AI-text detectors flag second-language writing incorrectly more often, so their results are never used to penalize authors.
Each year we check our screening tools for uneven results across countries, languages and institutions. We correct or withdraw any tool that shows unjustified bias.
Reviewer-finding software is used to widen the reviewer pool across regions, genders and career stages.
8. Technology for integrity screening
Submissions are screened with approved, confidential tools for plagiarism and duplicate publication, image manipulation, statistical inconsistencies, retracted or fabricated references, and signs of paper mills or manipulated peer review. A trained person reviews every result before any action is taken. Tools raise questions; people make decisions.




