TRONDHEIM, November 3, 2026 — New software sold online allows anyone to produce a ready-to-publish research paper in under two hours, and the resulting flood of studies is straining the peer-review system that guards scientific quality, researchers at the Norwegian University of Science and Technology (NTNU) have found.
The warning comes from statistician Andrea Riebler, a professor at NTNU’s Department of Mathematical Sciences, who traced the problem after citations of her own cancer research jumped from 70 in 2024 to more than 250 in 2025.
How a Citation Surge Led to a Discovery
Riebler published a 2017 paper with Professor Leonhard Held of the University of Zurich on forecasting future cancer burden, and she made the source code freely available with instructions on how to use it. The work was cited steadily for years, a sign it had made a mark among other researchers.
Then the numbers changed sharply. Almost all the new papers came from Chinese researchers, and most drew on the international Global Burden of Disease (GBD) project, a free public health database. The papers churned out an apparently endless series of forecasts on subjects ranging from stroke risk in adults over 20 to testicular cancer in young adults, falls among older people in China and colorectal cancer linked to low wholegrain consumption.
“Many of the papers had the same structure. Their titles were similar, with variations relating to different diseases,” Riebler said. “Some of the graphics were based on the same model we used, and their design was very similar to the one I used in my software package. That’s unlikely to be a coincidence.”
A Vendor in Guangzhou and a Trail of 700 Papers
Riebler, postdoctoral researcher Peter Methys Degen and Held followed the trail through more than 700 GBD papers. In one of them they found a reference to a commercial software package from China that includes instructions for doing exactly what had been done in many of the others.
In the Guangzhou region, a company sells software, AI-powered writing assistance and instructions that appear to be based on Riebler’s research. The team also identified two other Chinese companies selling similar packages for mass-producing papers from public health databases. Their findings appeared in the Journal of Clinical Epidemiology under the title “Paper mill or paper mine? A tentative answer to the sharp increase in research papers based on the Global Burden of Disease database.”
“We found evidence for decentralized mass production of papers based on secondary analysis of public health data,” Riebler said. “In their tools they also seem to use our work, which is how we found out about the software and how the research started.”
Sound Methods, Uncertain Value
The encouraging part, if there is one, is that the analyses themselves may be statistically sound. Riebler is careful not to claim the results are wrong. The concern is relevance and volume: whether each individual paper adds anything useful when the same public data are mined again and again.
“The analyses may be methodologically sound, but the development is worrying nonetheless,” she said. “The question is whether each individual paper is scientifically relevant and useful, particularly given the enormous volume of publications we’re now seeing.”
The papers rely on secondary analyses, meaning the data were collected by other researchers. Riebler questions whether every possible combination of data available in these databases produces meaningful research. She is not alone in noticing the pattern: the same exploitation applies to several other major free health databases established with good intentions, she said. “It’s frightening that this is happening.”
Missing Code Blocks Verification
For 99 percent of the papers examined, the authors did not provide their source code. That means the research cannot be reproduced, which violates a fundamental scientific principle meant to ensure accountability, transparency and quality.
Reproducibility matters to patients as much as to journals. When a study claims a disease trend or a risk forecast, other scientists need the code to check the calculation. Without it, a paper can look rigorous while remaining unverifiable.
“Sharing code should be mandatory,” Riebler said. The researchers are also calling for better monitoring of online platforms that sell ready-made research tools.
Editors Report 60 Percent Submission Spike
The pressure is landing on journals. Accountability in Research, which focuses on research integrity, has seen a 60 percent increase in submissions in 2026, and according to one of its editors, David Resnik, it is being bombarded, ironically, with papers that are probably AI-generated and deal with research misconduct.
“I’m very worried about this straining, breaking the back of the peer-review system,” Resnik told The Verge, the American technology news site that covered the NTNU findings.
Degen put the risk plainly. “It’s a huge burden on the peer-review system, which is already at the limit. There’s just too many papers being published and there’s not enough peer reviewers, and if the LLMs make it so much easier to mass produce papers, then this will reach a breaking point,” he told The Verge.
Riebler sits on the editorial boards of two scientific journals, and her inbox is filling with submissions too. She raised a question she does not pretend to answer: whether it matters if a paper is written by a language model or a human, as long as the content is correct. The problem, she said, is the sheer volume. “We’re already struggling to find peer reviewers, and the situation is getting progressively worse.”
What It Means for Readers of Health News
For anyone who follows cancer and medical research, the practical lesson is to look at where a study was published, whether its authors shared their data and code, and whether the finding is genuinely new. A forecast built from the same public database as hundreds of others deserves more scrutiny than a clinical trial in patients.
Peer review, the anonymous assessment of work by experts in the same field, is the mechanism that normally filters such papers. It depends on reviewers who are already stretched thin.
The NTNU team’s paper is published in the Journal of Clinical Epidemiology, and the researchers want code-sharing to become a condition of publication. Until then, patients should treat dramatic-sounding studies with caution and talk to their doctor before acting on any single paper.
Sources
Photo: Alex Knight / Unsplash (Unsplash License)
