1.5 lakh fake AI citations enter scientific records by 2025: Study | india news
Bengaluru: As a flood of AI-generated fake citations enter the scientific literature, a new study estimates nearly 1.5 lakh fabricated references entering the record by 2025, with the majority moving from preprints to peer-reviewed journals.This is the central finding of a large-scale study by researchers at Cornell University, UCLA and UC Berkeley, which analyzed 111 million citations in 2.5 million research papers published between 2020 and 2025 on arXiv, BioRxiv, SSRN and PubMed Central. The study, titled ‘LLM Hallucinations in the Wild’, tracked citations whose titles could not be verified against major academic databases including Semantic Scholar, OpenAlex and Google Scholar. By comparing trends after 2022 with a pre-ChatGPT error baseline, the researchers isolated the potential contribution of AI-generated hallucinations to the increase. The fastest growth began in mid-2024, about 18 months after ChatGPT’s public release, as AI tools evolved from writing assistants to citation-generation engines. The contamination is clearly not concentrated in the fraudulent papers. The researchers found that fake references are generally sparsely sprinkled into otherwise legitimate manuscripts, suggesting that many researchers are copying AI-generated citations without verifying them. Existing security measures are failing. Approximately 78.8% of duplicate citations passed arXiv moderation, and 85.3% of the duplicate citations in bioRxiv preprints subsequently published in PubMed Central-indexed journals made it into the final published versions. Researchers warn that the problem may now be self-reinforcing: As fabricated references embed themselves in open-access repositories and citation databases, future AI models trained on that corpus risk absorbing and reproducing the same hallucinations.Study in Lancet also warns In a separate study titled “Fabricated citations: an audit across 2.5 million biomedical papers,” published in The Lancet, researchers found a sharp increase in fabricated citations in biomedical research papers. The study, conducted by researchers at Columbia University and other institutions, analyzed biomedical papers published between 2023 and early 2026. It found 2,810 peer-reviewed papers contained more than 4,000 fabricated references. The audit found that the rate of fabricated references had increased dramatically over a three-year period. In 2023, approximately one in 2,828 papers contained at least one fabricated citation. By 2025, this figure had worsened to one in 458 papers, and by early 2026, it had reached one in 277 papers. One of the most striking examples cited in the study involved a 2025 paper in an open-access oncology journal on ureteral surgical techniques. The researchers found that 18 of the paper’s 30 verified references (60%) were fabricated. The authors link this increase partly to the widespread adoption of LLMs that are known to “hallucinate” fake citations. Warning that fabricated citations could compromise clinical guidelines and systematic reviews, researchers urged publishers to introduce automated reference verification systems before accepting papers for publication. The study said that about 98% of the affected papers did not face any publisher action at the time of the audit.
