1 of 29

Verifying the Accuracy of Citations Generated by AI Chatbots

September 29th, 2026

U-M Library Workshop

Kathryn Vanderboll, MLIS

Taubman Health Sciences Library

2 of 29

Housekeeping

  • This session will be recorded.
  • Feel free to put questions into the chat.
  • After this session, you will receive:
  • The recording
  • A copy of the slides

2

3 of 29

About me

  • Health sciences librarian
  • Frequently asked for help finding articles
  • Teach citation practices and citation management software
  • Regularly experiment with AI research tools

3

4 of 29

What AI tools are we talking about?

Multipurpose generative AI chatbot tools like:

  • ChatGPT
  • U-M GPT
  • Gemini
  • Claude
  • Copilot

4

5 of 29

What's the problem?

  • These chatbot tools can:
    • Suggest users cite works that may not exist at all
    • Format citations that may be incomplete or contain incorrect details
    • Incorrectly summarize real articles
    • Claim false information comes comes from real articles

5

6 of 29

A note on terminology

  • "Fake citations"
  • "Fabricated citations"
  • "Hallucinated citations"
  • "Counterfeit citations"
  • "Phantom citations"

A citation where the title does not match any real papers written by the listed author(s).

​

6

7 of 29

Another note on terminology

  • "Citation errors"

​

A citation where the title and author(s) match a real paper, but other citation details may be incorrect (journal, volume, issue, type of source, etc.)

​

7

8 of 29

Why does this happen?

Chatbot tools like ChatGPT are language prediction tools, suggesting answers based on the probability of words appearing next to each other.

Cabezas-Clavijo, Á., & Sidorenko-Bautista, P. (2026). Assessing the performance of 8 AI chatbots in bibliographic reference retrieval: Grok and DeepSeek outperform ChatGPT, but none are entirely accurate. Journal of Data and Information Science, 11(2), 342–354. https://doi.org/10.1515/jdis-2025-0326

Gibney, E. (2025). Can researchers stop AI making up citations? Nature, 645(8081), 569–570. https://doi.org/10.1038/d41586-025-02853-8

Linardon, J., Jarman, H. K., McClure, Z., Anderson, C., Liu, C., & Messer, M. (2025). Influence of topic familiarity and prompt specificity on citation fabrication in mental health research using large language models: Experimental study. JMIR Mental Health, 12(1), e80371. https://doi.org/10.2196/80371

Walters, W. H., & Wilder, E. I. (2023). Fabrication and errors in the bibliographic citations generated by ChatGPT. Scientific Reports, 13(1), 14045. https://doi.org/10.1038/s41598-023-41032-5

8

9 of 29

The impact…

  • Students may submit work with fake citations
  • Fake citations appear in preprints and journal submissions
  • Articles with fake citations are making it into the published literature

​

Camp, N. T., Bengtson, J. A., & Sandstrom, J. C. (2025). The citation catastrophe: Propagation of AI-generated counterfeit citations in scholarship. The Journal of Academic Librarianship, 51(4), 103065. https://doi.org/10.1016/j.acalib.2025.103065

Naddaf, M., & Quill, E. (2026). Hallucinated citations are polluting the scientific literature. What can be done? Nature, 652(8108), 26–29. https://doi.org/10.1038/d41586-026-00969-z

Topaz, M., Roguin, N., Gupta, P., Zhang, Z., & Peltonen, L.-M. (2026). Fabricated citations: An audit across 2.5 million biomedical papers. The Lancet, 407(10541), 1779–1781. https://doi.org/10.1016/S0140-6736(26)00603-3

Sevryugina, Y. V., & Vargas, D. (2026). Teaching research integrity through verification of AI-generated references: An activity for upper-level chemistry courses. Journal of Chemical Education, 103(5), 2610–2620. https://doi.org/10.1021/acs.jchemed.5c01620

Whitford, E. (2026, May 3). Journal submissions riddled with AI-created fake citations. Inside Higher Ed. https://www.insidehighered.com/news/faculty/books-publishing/2026/03/06/journal-submissions-riddled-ai-created-fake-citations

Many instances covered by Retraction Watch reporting (1, 2, 3, 4, 5, 6, 7, 8, 9, 10)

9

10 of 29

Reasons for the problem

  • Unintentional error or purposeful misconduct
  • Pressure to publish / finish assignments quickly
  • Lack of understanding of how to read and synthesize academic works
  • Uncritically trusting generative AI tools
  • Guidance struggling to keep up with rapid technological change
  • No easy solution for automated reference verification

Badalova, F., & Mayr, P. (2026). Detecting hallucinated and suspicious citations: What current tools can and cannot do (arXiv:2607.22693). arXiv. https://doi.org/10.48550/arXiv.2607.22693

​

10

11 of 29

Why isn't it easy?

  • Variations due to citation styles
  • Fake citations often copy elements of real citations
  • Languages other than English / translation variations
  • How to access the "complete" academic corpus?

​

Naddaf, M., & Quill, E. (2026). Hallucinated citations are polluting the scientific literature. What can be done? Nature, 652(8108), 26–29. https://doi.org/10.1038/d41586-026-00969-z

​

11

12 of 29

Who's verifying the citations?

For journal manuscripts:

  • The authors?
  • The journal editors?
  • The peer reviewers?
  • Each reader?

​

For student projects:

  • The student authors?
  • Student peer reviewers (if applicable)?
  • The faculty / TAs?

12

13 of 29

How to verify citations

13

14 of 29

When you have a DOI

  1. Click on the DOI of the citation
  2. Compare the elements of the citation to the journal's website.

​

14

15 of 29

Examples

15

16 of 29

If the DOIs are not hyperlinks

Example:

doi:10.1038/s41598-023-41032-5

Replace doi: with doi.org/

doi.org/10.1038/s41598-023-41032-5

16

17 of 29

If there's no DOI at all

Or the DOI takes you to a different article than you expected…

​

  1. Search for the article title on Google Scholar
  2. If title matches, follow the Google Scholar link to the journal's webpage

17

18 of 29

If no Google Scholar match

  1. Search for the author's website, CV, ORCiD, or Google Scholar profile
  2. Run an exact Google search for the title of the article (quotation marks)

18

19 of 29

My example (author search)

  • Searched for the authors (Fung, L., Richards, J., & Wenzel, J.)

19

20 of 29

Try an exact title search

  • Exact Google search for title of article, with quotation marks

20

21 of 29

Alternatives for finding literature

Instead of multipurpose genAI chatbots, try…

  • Library catalog or academic database searching
    • Find your librarian here

​

  • Specialized AI tools for finding academic literature ("AI discovery tools")

21

22 of 29

A note on AI discovery tools

  • Limited number of results retrieved
  • The search is a black box
    • Why are you getting these results?
    • Are they the most relevant?
    • What are you missing?
  • The search is not reproducible

22

23 of 29

A final word of caution

  • Generative AI tools can also make errors when summarizing or extracting information from real articles

​

  • Compare the original full text to any generative AI summaries

23

24 of 29

Alternatives for citation formatting

Citation management software can help you:

  • Keep track of the sources you're using for your research
  • Organize the sources you've found with folders or tags
  • Share those sources with collaborators
  • Read and annotate full text
  • Automatically create a draft bibliography and/or in-text citations/footnotes in a citation style of your choice.

24

25 of 29

What next?

25

26 of 29

Learn citation management software

26

27 of 29

For instructors

The library creates and maintains many asynchronous modules that can be added to your Canvas course.

  • GenAI and Library Research
  • Read a Scholarly Article
  • Introduction to Citation
  • Introduction to Zotero

27

28 of 29

Full bibliography

Badalova, F., & Mayr, P. (2026). Detecting hallucinated and suspicious citations: What current tools can and cannot do (arXiv:2607.22693). arXiv. https://doi.org/10.48550/arXiv.2607.22693

Cabezas-Clavijo, Á., & Sidorenko-Bautista, P. (2026). Assessing the performance of 8 AI chatbots in bibliographic reference retrieval: Grok and DeepSeek outperform ChatGPT, but none are entirely accurate. Journal of Data and Information Science, 11(2), 342–354. https://doi.org/10.1515/jdis-2025-0326

Camp, N. T., Bengtson, J. A., & Sandstrom, J. C. (2025). The citation catastrophe: Propagation of AI-generated counterfeit citations in scholarship. The Journal of Academic Librarianship, 51(4), 103065. https://doi.org/10.1016/j.acalib.2025.103065

Gibney, E. (2025). Can researchers stop AI making up citations? Nature, 645(8081), 569–570. https://doi.org/10.1038/d41586-025-02853-8

Linardon, J., Jarman, H. K., McClure, Z., Anderson, C., Liu, C., & Messer, M. (2025). Influence of topic familiarity and prompt specificity on citation fabrication in mental health research using large language models: Experimental study. JMIR Mental Health, 12(1), e80371. https://doi.org/10.2196/80371

Naddaf, M., & Quill, E. (2026). Hallucinated citations are polluting the scientific literature. What can be done? Nature, 652(8108), 26–29. https://doi.org/10.1038/d41586-026-00969-z

Sevryugina, Y. V., & Vargas, D. (2026). Teaching research integrity through verification of AI-generated references: An activity for upper-level chemistry courses. Journal of Chemical Education, 103(5), 2610–2620. https://doi.org/10.1021/acs.jchemed.5c01620

Topaz, M., Roguin, N., Gupta, P., Zhang, Z., & Peltonen, L.-M. (2026). Fabricated citations: An audit across 2.5 million biomedical papers. The Lancet, 407(10541), 1779–1781. https://doi.org/10.1016/S0140-6736(26)00603-3

Walters, W. H., & Wilder, E. I. (2023). Fabrication and errors in the bibliographic citations generated by ChatGPT. Scientific Reports, 13(1), 14045. https://doi.org/10.1038/s41598-023-41032-5

Whitford, E. (2026, May 3). Journal submissions riddled with AI-created fake citations. Inside Higher Ed. https://www.insidehighered.com/news/faculty/books-publishing/2026/03/06/journal-submissions-riddled-ai-created-fake-citations

​

​

28

29 of 29

Questions?

​

29