Between the Scylla of Prohibition and the Charybdis of Permissiveness: Journal Editorial Strategies in the Age of Generative AI Models

Abstract

Objective. To synthesise current editorial policies governing the use of generative artificial intelligence (AI) models in scholarly publishing and to identify unresolved issues requiring further guidance and evidence.Methods. A descriptive review of peer-reviewed publications (2023-2025) and openly accessible policies issued by publishers and journals was conducted.Results. In the absence of a unified international standard, major global and Russian publishers have issued role-specific guidance for authors, reviewers, and editors on interacting with generative AI (e. g., GPT-class models). Areas of emerging consensus include: AI systems are not recognised as authors; accountability for content resides exclusively with human contributors; and the use and role of AI must be transparently disclosed. Notable heterogeneity persists in the boundaries of permitted practices, ranging from non-binding «fair-use» recommendations to formal checklists and mandatory disclosure fields embedded in editorial management systems. Guidance is most developed for authors and editors, whereas rules for reviewers are comparatively sparse. Disciplinary variation is evident in both the permissiveness and specificity of recommended practices.Research gaps. There is no industry-wide consensus on acceptable uses of generative AI in research reporting or editorial workflows. Empirical evidence remains limited regarding the impact of generative AI on manuscript quality, the integrity and efficiency of peer review, and reader perception. Standards for provenance tracking and durable recording of AI-generated content are under-specified, and documented retractions explicitly involving AI-generated manuscripts are rare.Conclusions. While norms around authorship, responsibility, and disclosure are converging, operationalisation across journals and disciplines is inconsistent. Coordinated standard-setting and rigorous empirical studies are needed to evaluate risks and benefits and to support evidencebased policy.