Retracted Fish Nutrigenomics Paper Exposes Undeclared AI Use and Fake References

A review article that promised to map the future of fish nutrigenomics has been struck from the scientific record, and the reasons behind its retraction are drawing attention far beyond the aquaculture community. The Editor-in-Chief of Blue Biotechnology, an open-access journal published by Springer Nature, has formally retracted the paper titled Fish nutrigenomics: unravelling the genetic code for sustainable aquaculture and improved nutritional benefits, which originally appeared in the journal in October 2025. According to the retraction notice, published on 6 March 2026, the decision was driven by two intertwined concerns: the undeclared use of generative artificial intelligence in preparing the manuscript, and a pattern of self-citation recycling that the editor judged to be irrelevant to the paper’s stated topic. The case offers a vivid illustration of how the peer-review system is grappling with a new generation of research integrity problems that did not exist a few years ago.

The most striking finding documented in the retraction notice concerns the article’s reference list. The editor determined that the majority of the references cited in the paper appear to be non-existent. In practical terms, this means that a large share of the studies underpinning the review’s claims could not be traced to any real publication. Fabricated or hallucinated references of this kind have become a hallmark of text generated by large language models, which can produce fluent, plausible-looking citations that dissolve under scrutiny. When reviewers or readers attempt to locate the cited works, they find no matching titles, authors, journals or digital object identifiers. In this case, the density of apparently fictitious citations was judged sufficient to suggest that generative AI had been used in writing the manuscript, although that use was never disclosed by the authors.

The second pillar of the retraction concerned citation behaviour. The notice states that the authors numerously cited their own previous works, none of which were judged to be relevant to the published paper. Self-citation is not inherently improper; researchers frequently build on their own earlier findings, and in specialised fields a degree of self-referencing is expected. What troubled the editor in this instance was the combination of irrelevance and volume. Citation recycling of this sort can artificially inflate an author’s bibliometric profile, boost journal-level metrics and create a misleading impression of scholarly consensus around a topic. When the self-citations are also tangential to the subject matter, they add noise rather than substance, and they deprive readers of the genuinely relevant literature they would expect a review article to survey.

The retracted paper had been positioned as a synthesis of nutrigenomics research in fish, a field that sits at the intersection of genomics, nutrition and aquaculture science. Nutrigenomics, broadly defined, seeks to understand how dietary components interact with an organism’s genome, transcriptome, proteome and metabolome, and how those interactions shape growth, health and nutritional quality. Applied to farmed fish, the field promises to inform feed formulations that improve growth performance, disease resistance and the nutritional value of the final product for human consumers. With capture fisheries at or near their sustainable limits, aquaculture has become the fastest-growing animal food production sector worldwide, and the genetic optimisation of farmed species is a major research priority. A credible review of this landscape would have been a valuable resource for the community, which makes the integrity failure all the more consequential.

The authors of the retracted article were affiliated with several Indian research institutions, including the ICAR-Central Institute of Fisheries Education in Mumbai, the Sher-E-Kashmir University of Agricultural Sciences and Technology of Kashmir, the ICAR-Central Institute of Brackishwater Aquaculture in Chennai, and the College of Fisheries in Kishanganj under the Animal Science University in Patna. The retraction notice records that all authors disagree with the retraction. Disagreement is not unusual in retraction cases, and it does not by itself reverse an editorial decision. Under the guidelines that most publishers follow, the retraction notice remains permanently linked to the article, which stays online but is clearly watermarked as retracted so that readers who encounter it are not misled. The notice itself, published as an open-access document, has already recorded hundreds of accesses, reflecting the attention such cases now attract.

For the journal, the case underscores the difficulty of screening submissions in an era when generative AI tools can produce manuscripts that pass superficial readability checks. Peer review was designed to evaluate scientific merit, not to authenticate reference lists or detect machine-generated prose. Fabricated citations are often discovered only after publication, when readers attempt to follow up on the sources. Journals and publishers have been rolling out additional safeguards, including reference verification at submission, disclosure policies requiring authors to declare any use of AI in manuscript preparation, and clearer rules reserving authorship for humans who take responsibility for the work. The retraction notice in this case makes explicit that the undeclared nature of the suspected AI use was central to the decision, alongside the reliability problems it created.

The episode also highlights why retractions, though painful, are an essential part of the scientific ecosystem. The retracted review concerned a topic with real-world stakes: sustainable aquaculture and the nutritional improvement of farmed fish. Researchers, feed companies and policymakers who consult the literature on fish nutrigenomics rely on reviews to orient their work. A review built substantially on non-existent references cannot support that function, because its synthesis rests on evidence that does not exist. Leaving such a paper uncorrected would risk propagating false claims into downstream research, extension materials and even feed development decisions. By retracting the article and publishing a transparent notice, the journal protects the integrity of the record, even at the cost of acknowledging that a flawed publication slipped through.

The broader context is a rising tide of AI-related integrity cases across scientific publishing. Publishers have reported growing numbers of retractions in which fabricated references, nonsensical text or undisclosed AI assistance played a role. The phenomenon is not confined to any single discipline or region, and it has prompted international bodies to refine their guidance on authorship, disclosure and the responsible use of AI tools. At the same time, many researchers use AI legitimately, for language polishing, literature discovery or data analysis, provided that such use is transparent and that humans verify and take responsibility for the final content. The line drawn in this case was between undisclosed use that compromised the reliability of the manuscript and the disclosed, supervised use that publishers generally accept. The retraction notice suggests the authors crossed that line in ways that could not be remedied by correction.

For the aquaculture and nutrigenomics communities, the practical lesson is one of vigilance. Review articles shape research agendas, funding priorities and teaching materials, and their authority depends on the verifiability of every citation. Readers evaluating reviews in this field, as in any other, are increasingly advised to check that cited studies exist and genuinely support the claims attached to them, and to treat reference lists with the same critical attention once reserved for methods sections. The retracted paper’s promise, decoding the fish genome to make aquaculture more sustainable and more nutritious, remains a legitimate and active research goal pursued by many groups worldwide. What the retraction removes is not that goal, but a document that claimed to summarise it on unreliable foundations. The scientific record, corrected through the retraction notice now permanently attached to the article, is the mechanism by which that distinction is made clear to everyone who comes after.

Subject of Research: Retraction of a fish nutrigenomics review article over undeclared generative AI use and fabricated references

Article Title: Retraction Note: Fish nutrigenomics: unravelling the genetic code for sustainable aquaculture and improved nutritional benefits

Article References: Retraction Note: Fish nutrigenomics: unravelling the genetic code for sustainable aquaculture and improved nutritional benefits. (n.d.). https://doi.org/10.1186/s44315-026-00053-1

Image Credits: AI Generated

DOI: 10.1186/s44315-026-00053-1

Keywords: retraction, fish nutrigenomics, generative AI, research integrity, fabricated references, self-citation, aquaculture, Blue Biotechnology, Springer Nature, peer review, open access, sustainable aquaculture