AI-Generated Articles Falsely Attributed to Real Researchers in Academic Journal
A study by a Greek university professor discovered that 48 of 53 academic articles in a journal appear to have been AI-generated and falsely attributed to real researchers, including deceased academics and scholars at prominent institutions, undermining the credibility of scientific publishing.
What happened
A systematic use of generative AI created and published misleading scientific articles over several years in a specific academic journal, according to research by Professor Diomides Spinellis of the Athens University of Economics and Business.
In 2024, a completely fabricated article was discovered that was falsely attributed to Spinellis. Of 53 articles with minimal bibliographic references examined, 48 appear to have been produced by AI.
Many articles were falsely attributed to researchers at prestigious universities including Washington University, Texas A&M, University of California at Berkeley, USC Keck School of Medicine, HEC Montreal, University of Shanghai, George Mason University, DePaul University, and Penn State. In two cases, the falsely named authors were deceased when the articles were published.
Reported impact
The widespread distribution of artificially produced and falsely attributed scientific articles threatens the credibility of academic publications.
Index analysis
The research used automated tools to collect and analyze all articles in the journal, examining the number of bibliographic references, author affiliations, and email addresses. For detecting AI-written articles, researchers used a heuristic model based on the observation that AI assistants like ChatGPT struggle with creating reliable bibliographic references. A subset of articles was also examined manually for signs of AI authorship, and analysis was strengthened using the Turnitin AI detection tool.
Governance controls
The study recommends measures to prevent further occurrences, including strengthening author identification and reviewing practices for evaluating research output. Without effective countermeasures, the uncontrolled spread of AI publications risks seriously undermining trust in the scientific community.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.