A major new study has raised concerns over the growing use of artificial intelligence in scientific publishing after researchers discovered thousands of fake citations in peer-reviewed medical papers.
The peer-reviewed research letter, "Fabricated citations: an audit across 2.5 million biomedical papers," was published in The Lancet on May 7, 2026, found that nearly 3,000 biomedical research papers included references to scientific studies that did not exist. Researchers warned that the problem could damage trust in medical science and may even affect patient care if false information is used in clinical decisions.
Scientists from Columbia University School of Nursing analysed around 2.5 million biomedical papers published between January 2023 and February 2026 in PubMed Central's Open Access database. Using an AI-supported verification system, they examined more than 97 million references cited in academic papers.
The researchers identified 4,046 fabricated citations appearing across 2,810 published papers. According to the study, many of the fake references looked convincing and were difficult to spot because they closely resembled genuine scientific citations.
The team said the number of fabricated references increased sharply from mid-2024 onwards, a period that also saw a rapid rise in the use of AI writing tools by students, researchers and academics. The study estimated that the rate of fake citations increased from about four per 10,000 papers in 2023 to around 57 per 10,000 papers by early 2026.
Lead researcher Maxim Topaz said the findings highlighted a serious challenge for the scientific community. He warned that healthcare professionals, policymakers and researchers often rely heavily on published medical literature while making decisions about treatments, public health guidance and future studies.
Researchers involved in the project said some papers contained an unusually high number of fabricated references. In one case, a paper reportedly included 18 fake citations out of a total of 30 references. The study did not accuse all authors of deliberate misconduct but suggested that some false references may have been unintentionally generated through AI tools that create realistic-looking but inaccurate information.
The researchers urged publishers and academic institutions to strengthen fact-checking systems for references before papers are approved for publication. They also called for better monitoring of suspicious citations and recommended reviewing previously published papers to correct or retract articles where fake references may have influenced findings.
The study comes amid wider global debate over the role of artificial intelligence in education, journalism and scientific research. Experts have repeatedly warned that while AI tools can improve productivity, they may also generate false or misleading information if not carefully checked by humans.
