The citation catastrophe: Propagation of AI-generated counterfeit citations in scholarship

Abstract

Knowledge creation in the academy is built upon the research of previous scholars. This legacy scholarship is tracked in new papers through the use of citations. Citations of prior work in academic papers create a network of related scholarship that serves as a firm foundation for successive work. However, the authors of this paper document examples of “counterfeit citations”, almost certainly created by Large Language Models, which represent academic work which does not actually exist. The authors track the propagation of these example counterfeit citations in the literature and discuss the damage they cause, means of measuring comparative value in affected citation databases, and potential remediation of this problem moving forward.