Editorial policies for ethical use of artificial intelligence in rheumatology journals.

Abstract

The integration of artificial intelligence (AI) into research and publishing poses ethical challenges. Global editorial associations, including the International Committee of Medical Journal Editors (ICMJE), have updated their recommendations to safeguard transparency and accountability for AI use. The extent to which these recommendations have been enforced in specialist journals remains unknown. The aim of our study was to analyze the extent to which indexed rheumatology journals have adopted AI-related editorial policies, the scope of these policies, and their alignment with ICMJE recommendations. A total of 58 impact-factor rheumatology journals were analyzed in view of their AI-related editorial policies. Author instructions, ethics statements, and publisher guidelines were overviewed for (1) AI-related instructions; (2) alignment with ICMJE recommendations; (3) provisions regulating AI use by authors, peer reviewers, and editors; and (4) regulations concerning generative text, images, data, and analytical outputs. Of the 58 journals, 45 (77.6%) presented explicit AI editorial policies, while 13 (22.4%) lacked any AI-related guidance. The majority of journals (98.2%) endorsed ICMJE points on AI. High-impact journals-Nature Reviews Rheumatology, The Lancet Rheumatology, and Annals of the Rheumatic Diseases-demonstrated the most stringent governance, prohibiting AI use for analytical and creative roles, mandating detailed disclosure of AI tools and prompts, and banning AI use for peer review and editorial decision-making. The guidance on AI use by peer reviewers and editors was present in 45 journals (77.6%). Permissive uses of AI were largely confined to language editing under human supervision. Generative or substantive uses-such as producing figures, conceptual contents, or data-were broadly restricted. Indexed rheumatology journals demonstrate variable editorial policies of enforcing AI guidance. While the adoption of AI-related policies is mostly improving, a marked heterogeneity still exists, particularly between top-tier and lower-tier journals. Upgrades of editorial policies are warranted to safeguard the integrity and transparency of rheumatology sources.