HONG KONG (FN) — The University of Hong Kong has launched an investigation after a PhD candidate’s published academic paper was found to contain more than 20 fictitious references, reportedly generated using artificial intelligence. The incident has raised fresh concerns about the unchecked use of AI tools in scholarly research.
The paper, titled “Forty Years of Fertility Transition in Hong Kong,” was published in the journal China Population and Development Studies in October. It was co-authored by Bai Yiming, a doctoral student in the Department of Social Work and Social Administration at the University of Hong Kong. Bai specializes in demographic trends and social policy, with a focus on fertility and aging populations in East Asia. The corresponding author was Professor Paul Yip Siu-fai, Associate Dean of the Faculty of Social Sciences.
Social media users first flagged the issue, noting that many of the citations listed in the paper did not exist in academic databases. Subsequent reviews revealed that at least 20 of the 61 references were fabricated, prompting the university to initiate a formal inquiry into the matter.
Professor Yip, who served as the corresponding author, issued a public apology, acknowledging that the references were not properly verified before publication. He said the student had used AI tools to assist with formatting and citation generation, but the oversight was ultimately his responsibility.
The journal defended the paper’s core findings, stating that the conclusions remained valid despite the citation errors. However, it admitted to “mismatches and inaccuracies” in the reference list and said it was reviewing its editorial processes to prevent similar incidents.
Experts say the case highlights a growing problem in academia: AI tools such as ChatGPT and other large language models can produce plausible-looking but entirely fictional citations. These “hallucinated references” mimic real journal formats, include fabricated author names, and cite non-existent volumes — making them difficult to detect without manual verification.
Top universities are responding with new policies to address these risks. Harvard, Oxford, MIT, and Stanford now require researchers to disclose any use of AI in academic work. Most prohibit AI-generated citations unless they are manually verified, and emphasize that AI cannot replace human judgment or be credited as an author.
Similar incidents have surfaced globally. A recent wave of retractions followed the discovery of AI-generated mistranslations in scientific papers. In journalism, AI-written articles have included fabricated quotes and events, prompting outlets to revise editorial standards and label machine-generated content more clearly.
In response, journals and institutions are tightening oversight. Oxford University Press and Springer Nature have updated submission guidelines to require AI disclosure. The Committee on Publication Ethics (COPE) has issued recommendations urging editors to scrutinize AI-assisted manuscripts more closely.
The issue of academic integrity has also made headlines in the United States, where Claudine Gay resigned as president of Harvard University following allegations of plagiarism in her scholarly work. While her case did not involve AI, it underscores the heightened scrutiny facing academic leaders and the growing demand for transparency and accountability in research. Increasingly, artificial intelligence appears to be not only supplementing but also competing with traditional forms of misconduct like plagiarism — introducing new challenges for institutions striving to uphold academic standards.
The University of Hong Kong said it is reviewing its guidelines on AI use and will take appropriate action based on the outcome of the investigation. Bai Yiming has not publicly commented on the matter.
As AI tools become more embedded in academic workflows, scholars and institutions face a critical challenge: how to harness their benefits without compromising the credibility of research. The Hong Kong case may serve as a cautionary tale for universities worldwide.
