The scientific community is grappling with a growing crisis of fake citations, a phenomenon that has been exacerbated by the rise of AI tools. A recent study estimates that approximately 1.5 lakh fabricated references slipped into the scientific record in 2025, most of which originated from preprints and made their way into peer-reviewed journals. This alarming trend is not confined to fraudulent papers but is instead scattered across legitimate manuscripts, indicating that many researchers are copying AI-generated citations without proper verification. The study, conducted by researchers at Cornell University, UCLA, and UC Berkeley, analyzed 111 million citations across 2.5 million research papers published between 2020 and 2025 on arXiv, bioRxiv, SSRN, and PubMed Central. By comparing post-2022 trends against pre-ChatGPT error baselines, they isolated the likely contribution of AI-generated hallucinations to the surge. The steepest rise began around mid-2024, roughly 18 months after ChatGPT's public release, as AI tools evolved from writing assistants into citation-generation engines. Existing safeguards are failing, with nearly 78.8% of fake citations passing arXiv moderation, and 85.3% of hallucinated references making it into the final published versions on bioRxiv preprints. This contamination is not just a theoretical concern; it has real-world implications. A separate study published in The Lancet found a sharp rise in fabricated citations in biomedical research papers, with more than 4,000 fabricated references embedded across 2,810 peer-reviewed papers. The rate of fabricated references rose dramatically over the three-year period, from one in 2,828 papers in 2023 to one in 277 papers by early 2026. One striking example involved a 2025 paper in an open-access oncology journal, where 18 of the paper's 30 verified references (60%) were fabricated. The authors linked the surge partly to the widespread adoption of LLMs, which are known to "hallucinate" fake citations. This raises a deeper question: 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. The implications are profound, as fabricated citations could compromise clinical guidelines and systematic reviews, potentially leading to incorrect scientific conclusions and misinformed medical practices. Researchers are urging publishers to introduce automated reference verification systems before papers are accepted for publication, but the problem may now be self-reinforcing. As AI tools continue to evolve and become more integrated into the research process, the challenge of distinguishing between real and fake citations will only grow more complex. This crisis underscores the need for a comprehensive approach to address the issue, including better education and training for researchers, more robust verification systems, and a deeper understanding of the capabilities and limitations of AI in scientific citation generation.
