Study: rate of fabricated references in biomedical papers has grown 12x+ since 2023; in early 2026, one in 277 papers had at least one non-existent reference

The Growing Problem of Fabricated References in Scientific Research

A concerning trend has emerged within the realm of biomedical research: the rate of fabricated references in peer-reviewed papers has significantly escalated. Recent findings illustrate a striking intensification of this issue over the past few years.

Alarming Statistics

The research data reveals a disconcerting trajectory regarding research integrity:

• 2023: One in 2,828 papers included at least one fictitious reference.
• 2025: This rate increased to one in 458 papers.
• Early 2026: The figure reached alarming levels with one in 277 papers containing at least one non-existent reference.

This trend signifies a 12-fold increase in the occurrence of fabricated references within a span of less than three years.

Discovery and Investigation

Associate Professor Maxim Topaz at Columbia University's School of Nursing has been instrumental in documenting this troubling phenomenon. Researchers have detected patterns suggesting that artificial intelligence plays a critical role in the generation of these counterfeit citations, as authors increasingly depend on AI tools that can hallucinate or fabricate reference details.

Implications for Scientific Integrity

The rise of fake references represents a significant challenge to the credibility of biomedical literature. When researchers cite non-existent studies to bolster their assertions, it jeopardizes the very foundation of evidence-based medicine and scientific advancement. The ramifications are particularly severe because:

1. Fabricated references can mislead fellow researchers and healthcare professionals.
2. AI-generated hallucinations produce citations that may sound credible but are non-existent.
3. The swift acceleration of this issue indicates that it may be worsening rather than stabilizing.

Response from the Scientific Community

The revelation of this trend has sparked heightened scrutiny of published research and an urgent dialogue concerning the necessity for enhanced verification mechanisms and guidelines for authors utilizing AI tools in their research methodologies.