Authorship, Disclosures, and Conflicts of Interest in Papers Related to the Use of Artificial Intelligence

As artificial intelligence (AI), particularly generative AI, is being actively introduced and utilized in medical research and manuscript writing, new challenges are emerging in academic publishing, specifically regarding author attribution, transparency, and conflicts of interest. This review examines the current status of AI use in medical publishing by focusing on three key areas: author attribution, disclosure methods regarding AI usage, and conflicts of interest. The prevailing view to date is that AI cannot be recognized as an author because it lacks the capacity to assume the responsibility that is a core requirement of authorship. Therefore, AI contributions are generally disclosed in the acknowledgment section. As AI becomes more deeply involved in the analysis and manuscript writing processes, it is expected that discussions regarding the attribution of intellectual contributions and the boundaries between tools and contributors will become more active in the future. The transparency of reporting AI usage depends on how the AI contributes to the research or manuscript. While AI used for purposes such as grammar or spelling correction is often exempt from disclosure requirements, if AI contributes more substantially to the content of the paper-such as text generation, data analysis, or code development-it is necessary to report this by indicating such details explicitly within the paper. However, stances on the level of disclosure vary among journals, such as whether to reveal all details like model specifications or prompts. Nevertheless, when generative AI is used in the research itself, detailed reporting is increasingly emphasized to ensure the reproducibility and scientific validity of the findings. Furthermore, AI adds a new dimension to conflicts of interest. This includes financial interests related to AI development, data ownership, and potential biases inherent in training datasets and algorithms. Since these factors can influence research results in subtle ways, more transparent and comprehensive disclosure of conflicts of interest in AI-based research is crucial.