Scientific articles for journals and conferences are produced to order or via paid co-authorship by so-called “paper mills.” Marie Kunešová from the Faculty of Applied Sciences of the University of West Bohemia studied how these operations work and how they affect the scientific community.
The issue of predatory publishing agencies and the sale of co-authorship is being addressed more frequently around the world today, although it is almost exclusively connected with scientific journals. Marie Kunešová from the Faculty of Applied Sciences at the University of West Bohemia in Pilsen (FAV) was intrigued by the fact that no one had yet thoroughly investigated the impact of paper mills on conferences, even though, for example, in computer science, conferences are considered more important than journals. And that is precisely why the cyberneticist, in a joint research project with international colleagues, focused on this very area. In an interview for INFO.zcu, Marie Kunešová spoke not only about her work but also about how paper mills operate and their impact.
What criteria did you follow in your research, and how did you identify the articles you then analyzed?
We focused on paid co-authorship in the context of scientific conferences. We specifically targeted specific offers for articles on social media. For several months, we scoured sites like Facebook, LinkedIn, and Instagram, searching for keywords related to the sale of author positions. These publishing services don't really make a secret of it; enter the right phrases into a search engine, and you'll see plenty of offers pop up, including the article title, the author positions being sold (e.g., the third author position out of six), and the price. From these offers, we compiled a list of those focused on conferences, and then it was just a matter of checking whether each article had actually been published.
Is there any way to spot articles from such a "factory"? Do they have anything in common?
Most of them are low-quality, mass-produced texts, often plagiarized or describing experiments that never actually took place. Nevertheless, there are plenty of people who will pay to have their name on such an article, for example, because they want to advance in their careers and need more publications to do so. Unfortunately, it's not always obvious at first glance that an article is a fake, which is why they manage to get past peer review and into academic journals and conferences published by virtually every publisher.
What does this mean for prestigious conferences or conference proceedings?
As for the very top-tier conferences, I'm not at all worried that article mills will infiltrate them on a large scale. There's fierce competition and strict quality control at those events. These agencies prefer targeting venues with a high likelihood of acceptance. What is definitely happening, however, is that leading conferences are receiving an increasing number of submissions overall, leading to an overload for reviewers. As a result, they cannot devote as much attention to individual reviews as is needed.
Another problem is reputational. If word gets out that a certain publisher is publishing a large number of problematic conferences, not only does this make it an increasingly bigger target, but it also undermines trust in all of its titles. This personally concerns me quite a bit, and it is actually one of the reasons why I participated in this research to draw attention to the problem and encourage the publishers involved to start addressing it more systematically while there is still time.
How common are these unethical practices today? And are they on the rise with the advent of AI?
This is difficult to quantify, as detecting every single case is beyond human capacity. However, several studies have analyzed the prevalence of such problematic publications in specific fields. For example, in biomedical literature, which is generally considered one of the most affected fields, it is estimated that 3% of all published studies come from article mills; in some specialized fields, this figure exceeds 20%. In any case, however, research suggests that this is a very widespread problem that is constantly getting worse. Speaking for myself, what we found in our research is just the tip of the iceberg. And yes, the rise of generative AI plays a role in this.
What risks does paid co-authorship entail (aside from discrediting the client)?
It means, for example, the spread of false information and the deception of other scientists. You read about a new method in a scientific journal that relates to your research, and you start testing it yourself, only to find out after some time that it doesn't work at all because the original article was completely fabricated. This results in a huge amount of wasted time and, sometimes, wasted funding. And I probably don't even need to mention that when certain individuals fraudulently pad their CVs in this way, it is the honest scientists who suffer, as they have to put in significantly more effort to achieve the same level of productivity.
Is there any way to defend against the practices of article mills?
In today's scientific ecosystem, it's impossible to prevent this entirely. As long as there is pressure to publish as many scientific articles as possible and the associated reward system, there will, of course, be people who try to get around it by taking a shortcut. And where there is demand, supply will always follow. I can offer the following advice to potential readers of these articles: just as with the problem of predatory journals, it is important to be aware that this trend exists and to rely only on sources we trust, where we know that a rigorous peer-review process is in place. It's also a good idea to watch for potential signs that something is amiss with an article, whether it's unusual authorship, duplicate images, suspiciously good results, or references to articles that don't exist. Then there are sites like PubPeer, where people comment on other authors' publications. If we're unsure, it's a good idea to check there to see if anyone has already commented on the article in question
