Citation integrity in the age of AI: evaluating the risks of reference hallucination in maxillofacial literature

Abstract

The increasing adoption of large language models (LLMs) such as ChatGPT in academic writing has introduced both opportunities and risks. While these tools enhance productivity and accessibility, their reliability in generating accurate references remains uncertain. This short communication highlights the growing concern of ‘reference hallucination’, where AI-generated citations appear legitimate but are fabricated or contain significant metadata errors. Across all scientific disciplines, including oral and maxillofacial surgery (OMFS), where evidence-based practice is foundational, such inaccuracies can undermine academic integrity and clinical trust. This article summarizes common reference-related errors reported in literature and calls for heightened editorial vigilance, AI-literacy training, and the integration of real-time bibliographic tools. Responsible use of AI in scholarly publishing is essential to preserving the quality and credibility of surgical literature.