NEW DELHI: A review of 2.5 million biomedical research papers revealed that nearly 3,000 papers contained fake citations not found in scientific databases, highlighting a worrying trend in academic publishing as the use of artificial intelligence (AI) grows.
Papers were published between January 1, 2023, and February 18, 2026, in the open-access database PubMed Central, maintained by the US National Institutes of Health.
The results, published in a correspondence article in The Lancet, showed that out of 97.1 million verified references, researchers identified 4,046 fake citations across 2,810 papers, a growth of more than 12-fold since 2023, with the largest increase beginning in mid-2024, coinciding with the advent of AI-based writing tools.
“This discovery directly impacts patients as medical professionals make treatment decisions based on clinical guidelines,” said lead researcher Maxime Topaz, an associate professor at Columbia University’s School of Nursing and Data Science Institute.
“Medical professionals or clinical guideline developers have no way of knowing that the evidence they rely on does not exist. For example, one of the papers we reviewed had 18 out of 30 spurious references. Some of these citations are already cited by other papers and appear in the systematic reviews that inform clinical care,” Topaz said.
“Of the 97.1 million verified references, we identified 4,046 apocryphal references across 2,810 papers,” the authors wrote.
They said, “The manufacturing rate increased more than 12 times, from about four per 10,000 sheets of paper in 2023, to 51.3 per 10,000 sheets of paper in the fourth quarter of 2025, reaching 56.9 per 10,000 sheets of paper in early 2026.”
The researchers recommended that publishers check references in each submitted paper, and that indexing services add metadata to records, so that users can evaluate the accuracy of references.
They also urged major research integrity databases to create a dedicated category for fake references to enable systematic tracking and accountability.
The team called on publishers to retrospectively examine existing publications and issue corrections or retractions when false references harm a paper’s results.
This article was generated from an automated news feed without any modifications to the text.
