CoChat

Evaluating the Integrity of LLM-Generated Citations: Prevalence and Risks of Fabricated References in Scientific Literature

Open access
Picazo-Sanchez, Pablo, Ortiz-Martin, Lara · 2026 · Data

The DOI matches its Crossref record; 28 of the 28 references it cites were found in a registry or an index.

Abstract

Large Language Models have become important in our lives, and academia is not agnostic to this trend, offering tools like text rephrasing and summarisation. However, this integration raises significant concerns regarding the integrity of science. In this paper, we investigate hallucinations of LLMs when generating scientific references. Using nine LLMs, we generated a dataset of 74,196 BIBTEX references to quantify and analyse fabricated references, focusing on distinguishing between intrinsic and extrinsic hallucinations. Also, we extracted and analysed 127,063 references from 3541 published papers in 2023 to assess the prevalence of fake bibliographic data. Our manual verification process identified eight instances of fabricated references. While the overall rate is statistically low, the mere existence of fabricated content in the peer-reviewed literature is a critical integrity issue, demonstrating a vulnerability in current academic validation systems. The significance of our finding is not the statistical prevalence but rather the necessity for rigorous, human-validated processes to prevent the injection of spurious citations regardless of their source.

Overview

Integrity signals

Registry

Crossref · 10.3390/data11050122

Cites

28 of 28 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Verification

Claims the paper makes

We couldn’t retrieve this paper’s full text, so its claims and self-check weren’t run. The registry and citation checks still apply.

Citations

Cites

Show all 28 references

Research works printed but not linked: 4

  1. Kocmi, T., and Federmann, C. (2023, January 12–15). Large Language Models Are State-of-the-Art Evaluators of Translation Quality. Proceedings of the 24th Annual Conference of the European Association for Machine Translation, Tampere, Finland.
  2. Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A.N., Kaiser, Ł., and Polosukhin, I. (2017, January 4–9). Attention is all you need. Proceedings of the 2017 Conference of Advances in Neural Information Processing Systems, Long Beach, CA, USA.
  3. Huang, W., Liu, Y., Qin, H., Li, Y., Zhang, S., Liu, X., Magno, M., and QI, X. (2024, January 21–27). BiLLM: Pushing the Limit of Post-Training Quantization for LLMs. Proceedings of the International Conference on Machine Learning, Vienna, Austria.
  4. Elsevier (2026, March 26). The Use of Generative AI and AI-Assisted Technologies in Writing for Elsevier. Available online: https://www.elsevier.com/about/policies-and-standards/the-use-of-generative-ai-and-ai-assisted-technologies-in-writing-for-elsevier.

Cited by

See all 1

Metadata

Authors
Picazo-Sanchez, P., & Ortiz-Martin, L.
Affiliations
School of Information Technology, Halmstad University, 301 18 Halmstad, Sweden
Venue
Data
Publisher
MDPI AG
Published
2026-05-20
Volume
11
Issue
5
Pages
122
ISSN
2306-5729 (ISSN-L)
OpenAlex
W7161820716
Cited by
1
References
20
Citations by year
2026: 1
FWCI
5.77
Type
journal article
Language
en
Open access
Locations
  • Data (publishedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Topics
  • Social Sciences › Education › Academic integrity and plagiarism
  • Social Sciences › Library and Information Sciences › scientometrics and bibliometrics research
  • Social Sciences › Sociology and Political Science › Misinformation and Its Impacts
Keywords
LLM hallucination, citation integrity, large language models, peer review, scientific integrity, citation verification

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.