Many would-be whistleblowers write to us about papers with nonexistent references, possibly hallucinated by artificial intelligence. One reader recently alerted us to fake references in … an ethics journal. In an article about whistleblowing.
The paper, published in April in the Journal of Academic Ethics, explored “the whistleblowing experiences of individuals with disabilities in Ethiopian public educational institutions.”
Erja Moore, an independent researcher based in Finland, came across the article while looking into a whistleblowing case in that country. “I started reading this article and found some interesting references that I decided to read as well,” Moore told Retraction Watch. “To my surprise, those articles didn’t exist.”
The article joins a long list of publications flagged for fake references, which can be hallucinations generated by a large language model like ChatGPT.
Moore ended up analyzing all of the paper’s 29 references and found at least 19 of them appear to be fabricated, by her count. Eighteen of the Google Scholar links in the online reference section turn up empty. Moore dug in further, searching for the article titles, common works by the authors, and the journal’s volumes and issues to triple check whether some portion of the reference may have been incorrect rather than made up.
What she found in many cases were a nonexistent article title by authors who have written other papers on the reference’s topic; a similarly titled article in a completely different journal by different authors; or (and sometimes, and) a journal volume, issue and page number leading to a totally different article than the one in the references.
The Journal of Academic Ethics is published by Springer Nature. Eleven of the fabricated references cite papers in the Journal of Business Ethics — another Springer Nature title.
“On one hand this is hilarious that an ethics journal publishes this, but on the other hand it seems that this is a much bigger problem in publishing and we can’t really trust scientific articles any more,” Moore said.
Yelkal Mulualem Walle of the Department of Information Technology at the University of Gondar in Ethiopia, and the corresponding author of the article, told us he and his coauthors used ChatGPT to generate the references. “However, the research is real and used real data,” he said by email.
Coauthors Seyoum Tilahun Gedefaw at the university’s College of Education and Haregot Abreha Bezabih of the Federal Ethics and Anti-Corruption Commission in Addis Ababa did not respond to a request for comment.
Michael Stacey, Head of Communications, Journals for Springer Nature confirmed the publisher is aware of the concerns raised. “We take all such concerns about papers we have published extremely seriously and are now looking into the matter carefully,” he told us.
Hallucinated references in general “are an area we are actively exploring,” said Chris Graf, research integrity director for Springer Nature. “This is more complex than it may at first appear, as references can be detailed by authors in a variety of different ways, often do not include DOIs, and simple tools to identify hallucinated references can produce false positives.”
This isn’t the first time Moore has come across fake references. Last year she gave a presentation, summarized on her blog, on fake references in Finnish master’s theses.
“I have a very old-fashioned sense of justice and think it is a duty to report scientific misconduct,” Moore said.
Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on X or Bluesky, like us on Facebook, follow us on LinkedIn, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].