Expert: Guillaume Cabanac, Computer Scientist, University of Toulouse AI-generated references are polluting the scientific literature at scale. A Nature news feature by Miryam Naddaf and Elizabeth Quill documents how tens of thousands of 2025 papers may contain invalid citations fabricated by large language models. Computer scientist Guillaume Cabanac of the University of Toulouse discovered the problem first-hand when Google Scholar alerted him that his paper had been cited in the International Dental Journal — the citation was to a reference he had never written. Cabanac and colleagues propose automated reference verification built into submission workflows before peer review, plus integrity metadata on indexing services so researchers can flag suspicious citations automatically. The takeaway: an AI that generates plausible-looking references is not a reliable research assistant until publishers verify citations as part of the submission process. Source: https://www.nature.com/articles/d41586-026-00969-z