Ethical governance of artificial intelligence in scholarly publishing: a scoping review to inform policy development in medical universities

Abstract

While generative AI benefits scholarly publishing, it raises concerns about authorship, accountability, reproducibility, and ethics, necessitating clear institutional policies. Given the ethical responsibilities of medical universities as centers of research training and scientific integrity, developing transparent and responsible policies for AI use in scholarly publishing has become increasingly important. This review aims to map existing ethical guidance and policy initiatives to support the development of institutional AI policies within medical university publishing environments. This scoping review adhered to the Arksey and O’Malley (2005) framework and PRISMA-ScR reporting guidelines. Following the formulation of specific research questions, a comprehensive search strategy was developed and calibrated using keywords from related articles and relevant MeSH/Emtree terms. A systematic search of PubMed, Scopus, Web of Science, Embase, and the Cochrane Library was conducted up to May 2025. Key search concepts included Artificial Intelligence, Generative AI, Academic Publishing, Authorship, guidelines, and policy. Reference lists, key journals, and Google Scholar were hand-searched. A two-stage screening process applied relevance, English language, and full-text criteria. Data from the final included studies were extracted, charted, and collated to map the evidence and address the study’s objectives. Forty-six documents met the inclusion criteria, with a notable surge in publication activity after 2022, peaking between 2023 and 2025. The analysis identified a growing consensus on core recommendations for AI use. Key areas of consensus include a strict prohibition on AI authorship and the requirement for human verification of all AI-generated content, whereas the extent and mechanism of disclosing AI assistance remain contested, ranging from mandatory detailed reporting to voluntary minimal disclosure. For peer review, guidelines emphasize retaining full reviewer accountability, prohibiting AI from making core scholarly judgments (e.g., on methodology or recommendations), and ensuring confidentiality. Journals are urged to implement structured disclosure systems and clear sanctions. Critical gaps persist, including a lack of secure infrastructure, limited tools for detecting AI misuse, and concerns about equitable policy implementation for non-native English speakers. This scoping review highlights evolving yet inconsistent policy responses to AI in scholarly publishing. Findings advocate for a balanced, conditional approach to AI disclosure to ensure transparency without undue burden. Future work can broaden geographic scope and employ prospective methods to evaluate policy implementation. The overarching aim is to develop adaptable guidelines that maintain accountability and support responsible innovation.