Generative AI and Research Misconduct under South Korea’s National Research and Development Innovation Act

Purpose: This study examines how the use of generative artificial intelligence (AI) in research may be interpreted and regulated under South Korea’s National Research and Development Innovation Act and its Enforcement Decree. It also evaluates how AI-assisted research practices challenge the conceptual boundaries of the statutory categories of research misconduct. Methods: Through doctrinal legal analysis of Article 31 of the Act and Article 56 of the Enforcement Decree, common AI-assisted practices across the research cycle—design, literature review, data generation and analysis, manuscript writing, and the input of data into AI systems—were mapped to the Act’s misconduct taxonomy and related legal duties. Results: Generative AI may plausibly implicate fabrication, falsification, plagiarism, and improper authorship (Article 31(1)1), as well as improper ownership of research and development outcomes and breaches of security measures (Article 31(1)3–4). The analysis further indicates that AI use destabilizes categorical boundaries, as individual outputs may simultaneously involve invented content, distorted interpretation, and unattributed reproduction. Numerous research-integrity risks arise from failures in research processes, including nondisclosure, inadequate verification, weak provenance tracking, and irreproducible analysis pipelines. Conclusion: Legal and institutional responses should prioritize transparency across the research cycle and the development of auditable workflows, rather than focusing solely on sanctioning problematic outputs. Clear disclosure standards, verification obligations, reproducibility requirements, and stringent data-stewardship rules are necessary to address these emerging risks.