Columbia University nursing professor Maxim Topaz and colleagues reported evidence that fabricated references are embedded in biomedical literature at scale, raising new concerns about AI tools used in research writing. In a study published in The Lancet, the team audited nearly 2.5 million biomedical papers and 97 million citations indexed on PubMed Central. The researchers identified more than 4,000 fabricated references across nearly 3,000 papers, and Topaz said the rate rose sharply beginning in 2024 when AI tools became more widely used in research workflows. The study estimated the share of papers containing at least one fake reference increased from about one in 2,828 in 2023 to one in 458 by last year. For universities, publishers, and clinical research networks, the results intensify the need for citation verification protocols, AI-assisted writing guardrails, and stronger editorial checks—especially for medicine’s cumulative guideline and trial citation chain.