Evolving Journal Policies for Ethical Use of Generative AI in Scientific Publishing

Editorial Challenges in the Age of Generative AI

Abstract

The rapid integration of generative artificial intelligence (AI) tools like ChatGPT into academic writing has created an ethical grey zone for authors, reviewers, and editors. While these tools can enhance clarity and language, their unchecked use raises serious concerns about authorship, accountability, plagiarism, and data fabrication. Yet, the editorial community remains divided: Should AI-assisted writing be treated as ghost authorship? Can reviewers ethically use AI for peer review summaries? Do journals require disclosure only when AI is used by authors, or also when used during editorial workflows? Amid these unresolved tensions, leading medical journals and editorial bodies such as COPE and WAME are introducing policies that vary widely in clarity and enforcement. Some allow AI for grammar correction; others prohibit its use entirely in data interpretation. However, most journals lack consistent language on how to declare, limit, or audit AI use — leaving authors confused and reviewers unregulated. This short communication explores the gaps in current journal policies on AI use and calls for unified, transparent guidelines that balance innovation with integrity. We advocate for a three-tiered declaration system: (1) AI-assisted writing, (2) AI-generated content, and (3) AI use in peer review and editorial decision-making. As medical education and public health researchers increasingly turn to AI for manuscript preparation, the editorial community must act quickly to avoid reputational risks and uphold trust in the publishing process.