Cheating, Chatbots, and Change: A Bibliometric Mapping of Academic Integrity and Generative AI in Higher Education

Background and Aim: Generative artificial intelligence (GenAI) tools such as ChatGPT have rapidly entered higher education, reshaping how students write, learn, and are assessed, while intensifying concerns about plagiarism, contract cheating, and academic misconduct. Despite fast-growing commentary and empirical work, bibliometric evidence remains limited on how research at the intersection of academic integrity, GenAI, and higher education has organized itself in the post-ChatGPT period. This study maps publication volume, thematic structure, collaboration patterns, and early visibility indicators in this integrity-focused GenAI literature from 2023 to 2025. Materials and Methods: A bibliometric and science-mapping design was applied to records retrieved from The Lens (Lens.org) using a misconduct-salient query combining academic integrity terms, GenAI terms (e.g., generative AI, ChatGPT), higher education, and misconduct constructs (plagiarism, cheating, misconduct). Data were extracted on October 30, 2025. After screening and cleaning, 303 records were retained. Descriptive indicators were computed, and VOSviewer was used to generate keyword co-occurrence and co-authorship networks; influential publications were identified using Lens scholarly citation counts. Results: Publication output increased sharply from 2023, with substantial growth through 2025, although 2025 items had less time to accrue citations at extraction. Keyword mapping revealed four dominant clusters: integrity and misconduct, GenAI tool use and academic writing, student perceptions with population descriptors and selected discipline-linked terms, and ethics and governance. Collaboration patterns showed a small connected core among repeat contributors alongside a larger periphery of single or weakly connected authors. Source and citation indicators suggest that early visibility is anchored in a limited set of integrity-focused and education/educational technology outlets, with highly cited contributions dominated by early synthesis and framing work, alongside studies focused on assessment and detection. Country-level patterns should be interpreted cautiously because affiliation metadata were missing for many records; geographic findings, therefore, reflect the known-affiliation subset rather than a complete global distribution. Conclusion: In the first three years after ChatGPT’s release, integrity-focused GenAI research in higher education expanded rapidly and developed a recognizable thematic and network structure. The field remains strongly shaped by misconduct framings, but the map also shows branching toward student experience and governance, supporting the need for broader collaboration and evidence-informed institutional guidance.