Anima Books
books by holistic veterinarian Dr Christine King
Miscellany
The ABCs of reading medical research and review papers these days
Accept nothing. Believe no-one. Check everyting!
“When references point to nonexistent studies,
the evidence they claim to support is fictional.”
The Lancet just published a short ‘correspondence’ article which reveals a stunning trend in the biomedical literature.
Topaz M, Roguin N, Gupta P, et al. Fabricated citations: an audit across 2.5 million biomedical papers. Lancet, 2026; 407: 1779–1781.
Here are some highlights, with my emphasis added:
❝ Fabricated references (references whose claimed titles correspond to no existing publication) can arise from paper mill activity, intentional misconduct, or uncritical use of artificial intelligence (AI) writing tools.
Large language models (LLMs) generate plausible sounding but fictitious references, a well documented failure mode; previous studies estimate that 30–69% of LLM-generated references in biomedical contexts are fabricated.
These references are often correctly formatted, attributed to real researchers, and bear plausible publication dates, making them difficult to detect by conventional peer review. ❞
A fabrication rate — i.e., a fibbing rate — of even 30% is concerning; a fibbing rate of 69%?! Yikes!
Would you trust anyone who lied even just 30% of the time?
How would you know what was the truth and what was a lie if you didn’t check everything they say?
Accept nothing generated by “AI.”
Believe no-one using “AI” to write their publications.
Check everything generated by “AI.”
. . .
❝ We present findings from a reference-integrity audit of 2·5 million biomedical papers spanning 3 years, showing that fabricated references are embedded in the peer-reviewed literature at scale, and that the rate of fabrication is accelerating.
[…] In 2023, approximately one in 2,828 papers contained at least one fabricated reference. By 2025, this had risen to one in 458 and in the first 7 weeks of 2026, one in 277 papers had at least one fabricated reference. ❞
As you can see in the graph, the fabrication rate remained stable, at approximately 4 per 10,000 papers, throughout 2023 and the first half of 2024. But beginning in mid-2024, the rate rose sharply, perhaps even exponentially, reaching approximately 57 per 10,000 papers by early 2026.
Data collection for this study ended on 18 February, 2026, so the first quarter of 2026 is incomplete in this graph. The current figure is likely to be even higher.
What was going on in the time period depicted by this graph?
The authors noted that the sharp inflection in mid-2024 coincides with the expected publication lag following widespread LLM adoption. LLMs became broadly available in late 2022 and 2023. With the time from journal submission to publication averaging between 100 and 200 days, LLM-assisted papers would have begun appearing in PubMed Central from mid-2024 onward.
The authors considered two additional possibilities: (1) increased ‘paper mill’ activity, and (2) changes in journal indexing practices.
According to Wikipedia, a paper mill is “a business that produces poor quality or completely fraudulent journal papers that seem to resemble genuine research, as well as sell[ing] authorship on such papers.”
It’s worth noting that an increase in the number and output of paper mills is possible only with the aid of “AI.”
. . .
❝ A 2025 paper on ureteroileal anastomotic techniques in an open access oncology journal contained 18 (60%) fabricated references of 30 verified; each fabricated reference was tailored to the paper’s narrow surgical topic, attributed to real urologists, and bore claimed publication years of 2023 or 2024. ❞
This is just one example of how papers published from mid-2024 onward need to be read with caution.
Accept nothing generated by “AI.”
Believe no-one using “AI” to write their publications.
Check everything generated by “AI.”
. . .
❝ Beyond individual papers, we identified patterns consistent with paper mill activity: the same two authors appeared across 11 papers in a single surgical journal in 2025, with 15 fabricated references covering CRISPR diagnostics, AI-guided nanovaccines, and gut microbiome biomarkers, all sharing a core co-authorship pair. ❞
So, an impossible number and variety of papers published by the same author or small group of authors in a single journal or a single year is a strong indicator of paper mill-generated “AI” science slop.
. . .
❝ The fabricated references we identified were not obviously defective: topically specific, correctly formatted, attributed to real researchers, and beared [sic.] plausible publication dates. ❞
This sentence may be the most concerning, because it places the burden on the reader to check every reference— and who has time for that?!
Besides, it should not be up to the reader to verify every citation in a paper they’re reading!
Most, if not all, reputable journals now have statements and guidelines for authors on the use of “AI” in the preparation of a paper for publication in their journal. The publisher places the responsibility squarely on the author(s) to ensure the veracity of everything generated by “AI.”
And yet…
❝ Of the 2,810 affected papers [those with fictitious citations identified in this study], 98·4% had received no publisher action at the time of our audit. ❞
The authors recommend four actions, including this one:
❝ Third, publishers should retroactively screen existing publications and issue corrections or retractions when fabricated references compromise a paper’s conclusions. ❞
Until that becomes standard practice, it is still up to us, the readers, to…
Accept nothing generated by “AI.”
Believe no-one using “AI” to write their publications.
Check everything generated by “AI.”
. . .
The penultimate sentence in this short ‘correspondence’ paper sums up the problem perfectly:
When references point to nonexistent studies,
the evidence they claim to support is fictional.
Their final sentence states that “Routine automated verification can close this gap before fabricated references reach the published record.”
Yes, indeed.
Authors, editors, and publishers: it is up to you, not me, to ensure that what I am reading in your journal is reliable!
. . .
ADDENDUM 27 May, 2026:
This article begins:
"A network of fake academic journals masquerading as legitimate publications has published more than a hundred AI-generated papers in recent months, in some cases using the names of real professors from top universities without their knowledge."
As the subhead says, the exploited professors have called this situation a fiasco and "a warning about the future of scientific knowledge."
Indeed.
. . .
© Christine M. King, 2026. All rights reserved.
First published on 09 May, 2026.
Updated on 27 May, 2026.