AI Audit Uncovers Thousands of Fabricated Citations in Medical Literature

A study published in The Lancet on May 7, 2026, reports that an AI-assisted audit identified nearly 3,000 peer-reviewed medical papers containing fabricated citations that do not correspond to any existing publication in scientific databases.

Key Findings

Researchers at Columbia University School of Nursing developed an automated verification system to scan 2.5 million papers in PubMed Central's Open Access database, published between January 1, 2023, and February 18, 2026.

The system verified 97.1 million references, identifying 4,046 fake citations across 2,810 papers.

The rate of fake citations increased more than 12-fold since 2023, with the sharpest rise beginning in mid-2024.

Study Details

The research was led by Maxim Topaz, PhD, an associate professor at Columbia University's School of Nursing and Data Science Institute. Co-authors include Nir Roguin, MD (Tel Aviv Sourasky Medical Center, Ben-Gurion University); Pallavi Gupta, PhD; Zhihong Zhang, PhD (Columbia University); and Laura-Maria Peltonen, PhD (University of Eastern Finland, Wellbeing Services County of North Savo, University of Turku).

Fabricated references are citations that do not correspond to any existing publication in scientific databases. The increase in these citations coincided with the increased use of AI writing tools.

Responses and Recommendations

Maxim Topaz stated that medical professionals and guideline developers cannot verify whether cited evidence exists. He noted that one paper reviewed contained 18 fake references out of 30, and some of those citations already appear in other papers and systematic reviews that inform clinical care.

The study authors recommend:

• Publishers verify references with each paper submission.
• Indexing services add metadata to records to allow users to assess reference accuracy.
• Major research integrity databases establish a dedicated category for fake references to enable systematic tracking and accountability.
• Publishers retroactively screen existing publications and issue corrections or retractions where fake references compromise a paper's conclusions.

As of the audit date, 98.4% of affected papers had not received any publisher action.

Commentary

In an accompanying commentary, Howard Bauchner, MD (Boston University) and Frederick P. Rivara, MD (University of Washington) stated that public trust in science is declining globally and that there is a need for renewed efforts to enhance research integrity. They noted that authors must be accountable for the entire manuscript, including references.