A team led by Columbia University School of Nursing researcher Maxim Topaz reported evidence that fabricated references are embedded in biomedical publications at scale. In work published in The Lancet, the researchers audited nearly 2.5 million papers and 97 million PubMed Central citations, finding more than 4,000 fake references across nearly 3,000 papers. The authors say the rate climbed more than 12-fold over three years, with the steepest acceleration occurring around 2024 as AI tools became more widely used in research workflows. The study’s findings highlight a specific mechanism—AI prioritizing language patterns over accuracy—that can compromise downstream work. For universities and journals, the development raises immediate questions about reference verification processes, editorial screening, and how faculty and students should validate AI-assisted citations before publication.