A contemporary find out about signifies an alarming determine: greater than 250,000 medical articles associated with most cancers can have been invented between 1999 and 2024. This manufacturing is accelerating and endangering fair medical manufacturing.
The manufacturing of information thru medical analysis ends up in sturdy pageant between groups and people, wherein a newsletter in a prestigious magazine can trade the trajectory of a profession. Even supposing many query the present regulations of this pageant, the analysis of the standard of a researcher is basically in response to the collection of his publications, on their have an effect on – measured via the quantity of citations they generate – and at the recognition of the journals wherein they’re printed. The significance of those signs in acquiring scarce investment and advancing particular person careers contributes to encouraging conduct opposite to medical integrity, comparable to using fraudulent practices.
This context specifically preferred the emergence and robust expansion of organizations specialised within the sale of pretend medical articles, “paper factories” or “article factories”. The latter are suspected of getting produced hundreds of articles over the last ten years, compromising complete swathes of the medical literature. In our find out about, printed within the British Scientific Magazine (BMJ) in January 2026, we estimate that greater than 250,000 medical articles associated with most cancers can have been produced between 1999 and 2024.
Whilst those articles represented not up to 1% of annual medical publications in 1999, their price is now 15% of the content material produced each and every 12 months. Most cancers analysis is at risk: faux publications are spreading, and this downside is predicted to accentuate.
Manufacturing on an commercial scale
“Scientific article factories” produce and promote faux medical articles in virtually commercial amounts. They even undertake vintage advertising and marketing ways, promoting on-line and providing their shoppers the chance to make a choice their position within the checklist of authors of the edited article (the primary and closing positions are continuously perceived as extra prestigious), in addition to the extent of recognition of the magazine wherein the object can be printed. Analysis has proven that the cost of this carrier can range from a number of hundred to a number of thousand euros, and a few suspect that some “article factories” may also supply “after-sales service”, for instance offering corrections or responses to readers’ feedback after newsletter, on writer websites or on collaboration platforms comparable to PubPeer.
Instance of an “item factory” advert posted on Fb on February 19, 2026. Delicate knowledge has been hidden. Equipped via the writer
The collection of publications attributed to “item factories” exploded within the early 2010s, confirming the life of a large-scale fraud machine. The expansion of those organizations is continuously introduced as a result of the so-called “Publish or perish” tradition, which draws a clientele of PhD scholars, researchers and suffering clinicians for whom newsletter has change into a demand for get entry to to some extent, activity or promotion.
This phenomenon may be strengthened via the life of middleman brokers and arranged networks that transcend the easy manufacturing of manuscripts and intrude via bypassing and dashing up editorial and publishing processes (faux articles will also be permitted and printed a lot sooner than original articles). This collusion between unscrupulous manufacturers and publishers contributes to a vital building up within the price of newsletter of pretend articles, to the level that it will probably very much exceed the velocity of newsletter of original articles.
Factories derive their productiveness from predefined editorial templates, continuously reusing fragments of textual content and photographs from their earlier productions. This system preferred the newsletter of articles presenting similarities in shape, together with the similar turns of word, the similar experimental design, and the similar methodological or stylistic mistakes within the biomedical literature, specifically in most cancers analysis.
Those strains make it imaginable to practice their path and systematically determine their manufacturing, as when it comes to manipulated pictures or “tortured sentences” (bad reformulations of technical phrases, for instance “breast danger” as a substitute of “breast cancer”).
A easy however efficient detection means
Our battle towards medical fraud started in 2024 when Baptist Scancar, the writer of this newsletter, then a Grasp of Knowledge Science pupil, went to Australia to paintings on medical fraud with Jennifer A Byrne (Professor of Oncology on the College of Sydney) and Adrian Barnett (Professor of Statistics on the Queensland College of Era). For years, Jennifer has noticed the abuse of her self-discipline, molecular oncology, infected on a massive scale via fraudulent publications, unbeknownst to communities or analysis establishments. The function of this collaboration used to be to create one way which may be generalized to very large amounts of articles to locate the manufacturing of “article factories” within the box of most cancers, to measure its extent and warn of the underlying downside.
The remark of sturdy stylistic similarities within the identify and summary of the faux articles led the workforce to change to research strategies interested in those portions of the articles, which can be additionally freely to be had on on-line medical distribution platforms comparable to PubMed. The function is to make use of a man-made intelligence set of rules to differentiate original articles from faux ones via figuring out commonplace patterns within the checklist of articles recognized as faux via Retraction Watch. This means is in the long run a success, attaining very top identity efficiency (9 out of 10 articles are smartly categorized via the set of rules).
After this primary segment, the challenge endured in France from 2025. The workforce is finished via David Causeur, additionally the writer of this newsletter. The software has been advanced with an in depth research of identity mistakes, and the method has been advanced. The style is then used to research all cancer-related publications since 1999, i.e. greater than 2.5 million articles to be had within the PubMed database, and the effects are alarming. About 250,000 articles, just about 10% of the research, had been reported to be textually very similar to merchandise manufacturing unit manufacturing. Their quantity larger from 238 in 1999 to greater than 26,000 in 2020. This growth does no longer spare even probably the most prestigious journals (best 10 newspapers), the place the percentage of printed articles may also exceed 10% in 2022.
The geographic distribution of authors highlights the numerous predominance of China, with just about 180,000 articles recognized, a long way forward of america (10,500 articles) and Japan (6,500 articles). The questionable publications are present in journals from quite a lot of publishers, overlaying all kinds of most cancers and maximum analysis subjects.
The most important second for medical analysis
The overpowering presence of counterfeit pieces within the most cancers box items a number of primary issues.
To start with, the sharing of medical wisdom nowadays is closely polluted with false knowledge, and analysis stakeholders are suffering to put in force corrective measures. The level of those primary breaches of medical integrity should additionally result in mirrored image at the weight given to the quantity of publications within the analysis of analysis initiatives and analysis groups themselves. Implicit validation of false medical findings thru their newsletter, every now and then in prestigious journals, compromises investment allocation processes and paves the way in which for its cascading unfold thru citations.
Additionally, elementary most cancers analysis, which is a fascinating goal of “article factories”, precedes the improvement of healing therapies, the effectiveness of which is compromised via large-scale fraud, to the detriment of sufferers.
Fresh trends in generative language fashions, comparable to ChatGPT, threaten to make it more difficult to locate fraudulent content material and will building up the productiveness of those organizations tenfold. Via analogy with doping in sports activities, it will have to be feared that this cat-and-mouse sport between detectors and fraudsters is not going to result in the eradication of the issue, however to the escalation of fraud methods.
Then again, public analysis analysis insurance policies can treatment this via lowering the force on researchers to post. It’s pressing to present extra space to the very important high quality of medical productions of their analysis, sooner than the variation between original and fictional content material turns into unattainable.
