EGU journals have established policies regarding the use of AI in manuscript preparation and handling, with guidelines provided for authors, editors, and referees. However, as AI tools continue to develop and become more widely used, the boundaries between appropriate and inappropriate use may not always be straightforward, and our understanding of ethical AI use in scholarly publishing will continue to evolve. This discussion will explore what constitutes inappropriate or unethical AI use in the publishing process and how potential cases may be detected and addressed. We will discuss how to distinguish acceptable from problematic use by authors, referees, and editors, how concerns about non-compliance should be evaluated, and what responses are appropriate when policies are not followed. The goal is to develop a shared understanding of how to handle suspected cases in a fair, transparent, and consistent manner, while ensuring that EGU policies remain adaptable as AI capabilities and publishing practices continue to evolve.