Mastodawn

And to finish off with a quote:

"Since LLMs generate their output by predicting word sequences, there is a sense in which every GenAI citation is hallucinated because it has been made up! However, people usually do not refer to these word predictions as hallucinated when they happen to exactly match existing citations."
🤯 😂

#ResearchEthics #ResearchIntegrity

Show thread

Jan R. Boehnke 4d ago

There are important side notes.

The discussion points to the general problem of inaccurate citations in the scientific literature, e.g., via research
https://link.springer.com/article/10.1186/s41073-025-00173-z
and proposals to track them
https://link.springer.com/article/10.1186/s41073-020-00099-8

And it alerts to 'stealth corrections', “post-publications changes, without providing any indication that the publication was temporarily or permanently altered” (from https://onlinelibrary.wiley.com/doi/10.1002/leap.1660)

#StealthCorrection #AcademicPublishing #ResearchEthics #ResearchIntegrity

Systematic review and meta-analysis of quotation inaccuracy in medicine - Research Integrity and Peer Review

Background Quotations are crucial to science but have been shown to be often inaccurate. Quotation errors, that is, a reference not supporting the authors’ claim, may still be a significant issue in scientific medical writing. This study aimed to examine the quotation error rate and trends over time in the medical literature. Methods A systematic search of PubMed, Web of Science, and reference lists for quotation error studies in medicine and without date or language restrictions identified 46 studies analyzing 32,000 quotations/references. Literature search, data extraction, and risk of bias assessments were performed independently by two raters. Random-effects meta-analyses and meta-regression were used to analyze error rates and trends (protocol pre-registered on OSF). Results 16.9% (95% CI: 14.1%-20.0%) of quotations were incorrect, with approximately half classified as major errors (8.0% [95% CI: 6.4%-10.0%]). Heterogeneity was high, and Egger’s test for small study effects remained negative throughout. Meta-regression showed no significant improvement in quotation accuracy over recent years (slope: -0.002 [95% CI: -0.03 to 0.02], p = 0.85). Neither risk of bias, nor the number of references were statistically significantly associated with total error rate, but journal impact factor was: Spearman’s ρ = –0.253 (p = 0.043, binomial test, N = 25). Conclusions Quotation errors remain a problem in the medical literature, with no improvement over time. Addressing this issue requires concerted efforts to improve scholarly practices and editorial processes.

SpringerLink

Jan R. Boehnke 4d ago

This paper presents a principled argument within the U.S.’ misconduct regulations that hallucinated citations may be a form of data fabrication and therefore count as research misconduct.
https://www.tandfonline.com/doi/full/10.1080/08989621.2026.2645390

#ResearchEthics #ResearchIntegrity

Jan R. Boehnke May 22

RE: https://fediscience.org/@satrevik/116612366859411834

Really interesting question.

I brought something similar last month to our #Postgraduate group with this paper as an example:
https://www.tandfonline.com/doi/full/10.1080/03075079.2026.2633517

This led to a good discussion about #AcademicCronyism. Whatever one thinks about the items in the poll, the paper offers some shades of grey regarding such actions.

#PublicationEthics #AcademicPublishing #ResearchEthics #COPE

LeidenForce May 20

A new study warns that automated writing tools may amplify plagiarism in scientific papers.

As AI speeds up publishing, concerns grow about integrity and originality in research outputs.

🔗 https://phys.org/news/2026-04-plagiarized-automated.html

#AI #ScientificIntegrity #ResearchEthics #Publishing #OpenScience

Plagiarized research passed automated tests, and I detected it—but only because it copied my work

Earlier this year, I published a paper on the ethics of researching military populations. The core argument was straightforward: the standard rules researchers follow to protect participants—for example, informed consent and voluntary participation—don't work the same in an institution built on hierarchy and obedience.

Phys.org

Andy Tseng May 15

😂 #ResearchEthics #ResearchSky #AcademicSky #AI #GenAI

RE: https://bsky.app/profile/did:plc:wld6fad6xsm4tz4kfkoikun2/post/3mlvxiv64cc2x

Eugene May 14

❗Update on Plagiarism and Falsification in Scientific Publications

Today, I officially submitted requests for the retraction of articles to the editorial boards of the journals "Land Reclamation and Water Management" and "Bulletin of NUWM".

❗Key facts of misconduct:
🔹 Plagiarism: Unauthorized use of my original relief map (2015) without citation.
🔹 Falsification: Intentional removal (retouching) of the author’s stamp with coordinates and date-time on site photographs.
🔹 Duplication: Publication of identical material in 2021 and 2024.

Next steps:
In the event of attempts to covertly remove the articles without an official retraction notice, relevant complaints will be forwarded directly to @crossref and the international members of the editorial boards.

#OpenScience #ResearchMisconduct #AcademicMastodon #ScientificIntegrity #Hydrology #Geoscience #GIS #FediScience #Plagiarism #SvystunovaGully #COPE #Crossref #DataIntegrity #ImageManipulation #AcademicChatter #ResearchEthics #EarthScience #HigherEducation

Eugene May 13

❗ While preparing a literature review for my monograph on the hydrogeochemical impact of mine waters, I discovered the unauthorized use of my original materials in articles published in 2021 and 2024. Both publications were prepared with the primary involvement of staff from the Institute of Water Problems and Land Reclamation of the National Academy of Agricultural Sciences of Ukraine (IWPLR NAASU).

An analysis revealed facts that go beyond a simple copyright violation:
🔹 Data Falsification: Original photo metadata (date, time, GPS) was intentionally retouched to conceal their 2015 origin and authorship.
🔹 Methodological Anachronism: Using 2015 data for 2024 modeling renders results invalid by ignoring the major infrastructure overhaul launched in early 2025.
🔹 Institutional Failure: This misconduct by state academy staff signals a systemic collapse of research quality control and ethical oversight within the institutions.

#ScientificMisconduct #SvystunovaGully #ResearchEthics #DataIntegrity #Plagiarism

Serhii Nazarovets May 12

In this presentation, I explore the idea of "AI plagiarism" not as a simple violation, but as a mismatch between new modes of text production and outdated systems for assessing knowledge and contribution.

https://doi.org/10.13140/RG.2.2.25560.87049