Phantom references and the peer review crisis: how artificial intelligence is testing the resilience of scientific publications

Abstract

The article examines the critical vulnerability of the traditional blind peer-review system to the challenges posed by the rapid development of generative tools. Based on a real-life case — the discovery of a completely fabricated bibliography in a submitted manuscript — the basic mechanisms of creating false references are analyzed: anachronisms, "Frankensteinization", professional bias. The evolution of the threat is traced — the transition from obvious errors of early algorithms to deep semantic hallucinations of modern RAG-based search engines capable of generating perfectly formatted, but conceptually empty texts from real databases. To protect the publishing process, an updated editorial control algorithm is proposed, which involves mandatory verification of digital object identifiers (DOI) and clear declaration of the algorithms used by the authors. The necessity and irreversibility of the transition to the Open Science Framework paradigm is emphasized, where textual material is considered only as a supporting document to a verified array of primary data sets, open code, and deposited collection samples.

Cite this article: 

Kharchenko V.O., Korneyev V.O., Filimonova N.S. Phantom references and the peer review crisis: how artificial intelligence is testing the resilience of scientific publications. Visn. Nac. Akad. Nauk Ukr. 2026. (6): 78—84. https://doi.org/10.15407/visn2026.06.078