Thousands of medical papers found citing fake scientific research
text_fields
Thousands of biomedical research papers were found to contain citations to scientific studies that do not exist.
A new study published in The Lancet has raised concerns about the growing use of artificial intelligence in academic publishing.
The peer-reviewed research letter, titled “Fabricated citations: an audit across 2.5 million biomedical papers,” analysed around 2.5 million papers published between January 2023 and February 2026 in PubMed Central’s Open Access database.
Researchers from Columbia University School of Nursing examined more than 97 million references using an AI-supported verification system and identified 4,046 fabricated citations across 2,810 published papers.
According to the study, many of the fake references closely resembled legitimate scientific citations, making them difficult to detect.
The researchers said the number of fabricated citations rose sharply from mid-2024 onwards, coinciding with increased use of AI writing tools among researchers and students. The rate of fake references reportedly climbed from around four per 10,000 papers in 2023 to 57 per 10,000 papers by early 2026.
Lead researcher Maxim Topaz warned that the issue could affect public trust in medical science and potentially influence healthcare decisions if inaccurate information is relied upon in clinical practice or policymaking.
The study noted that some papers contained large numbers of fabricated references. In one reported case, 18 out of 30 citations in a paper were found to be fake.
Researchers said not all cases necessarily involved deliberate misconduct. Some fabricated citations may have been unintentionally generated by AI tools capable of producing realistic but inaccurate information.
The team called on publishers and academic institutions to strengthen fact-checking systems for references before approving papers for publication. They also recommended reviewing previously published research to identify and correct papers containing false citations.
The findings add to growing global concerns about the risks associated with unchecked use of artificial intelligence in research, education, and media.