Answers are grounded in this paper and cite their sources.
Generative artificial intelligence (AI) tools have entered manuscript preparation, peer review, and editorial decision making faster than journals have built policy to govern them, raising ethical questions about fairness, accountability, confidentiality, and honesty. This scoping review maps the published literature on AI and research integrity in scholarly publishing and analyzes its ethical implications. We searched Semantic Scholar, OpenAlex, and CrossRef for records published between 2000 and 2026 and added a supplementary targeted search. We included 77 studies (37 from the database search, 40 from the supplementary search). Empirical or benchmark studies form the largest group (27 studies, 35%), followed by editorials and commentaries (20, 26%). Detection tools show inconsistent accuracy and documented bias against non-native English writing. Authorship guidance converges on excluding AI systems as authors while permitting disclosed assistive use. Guidance for AI used as a writing tool is less developed than guidance for studies that evaluate AI systems. Building on this evidence, we argue that detector output alone is an unfair basis for sanctioning authors, that responsibility for AI-assisted text cannot be delegated to a tool, that reviewers’ confidentiality duties require institutional support, and that disclosure requirements should be proportionate and paired with fair verification procedures. The evidence base rests largely on opinion and small studies, and the effectiveness of verification mechanisms remains untested.
No takes yet. Share an insight, caveat, or question.
Forkuo et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: