Generative AI ethics guidelines for academic writing: A content analysis of international publisher policies and a roadmap for the Indonesian context

Abstract

The rapid rise of generative artificial intelligence (AI) tools such as ChatGPT and Grammarly has affected academic writing and literacy—a core concern in applied linguistics that encompasses genre conventions, authorial voice, and scholarly textual practices. While these tools enhance linguistic quality and efficiency, they pose ethical risks, including inadvertent plagiarism, reduced originality, and algorithmic bias. This study employs discourse analysis to examine AI ethics policies from five leading international publishers—Elsevier, Springer Nature, Wiley, Taylor & Francis, and SAGE—and focuses on how these policies linguistically construct norms of AI use through modality, lexical choice, and author accountability. Results reveal convergences (e.g., prohibition of AI as author, mandatory disclosure, human accountability) and divergences (e.g., Wiley’s documentation requirements, Springer’s confidentiality rules for reviewers, Taylor & Francis’ explicit mention of generative tools, and SAGE’s broader scope). Based on these findings, we propose a roadmap for standardized AI ethics guidelines for Indonesian journals, including: (1) tiered AI disclosure policies distinguishing language polishing (simple acknowledgment) from generative content creation (detailed documentation), and (2) institutional oversight committees comprising editors, ethicists, and AI literacy experts to review borderline cases. This roadmap also advances applied linguistics research on GenAI-textual patterns, including corpus-based detection of AI-generated academic text in non-native English contexts. Recommendations include policy harmonization, editor training, and disclosure templates to ensure ethical integration of AI in scholarly publishing.