Scientific Publishers Combat Fraudulent Papermills
Publishers are deploying new integrity teams and AI tools to fight the surge in fraudulent papers from commercial papermills, a problem accelerated by
Scientific publishers are building specialized teams and developing new tools to combat a surge in fraudulent papers from commercial papermills. The problem, accelerated by the use of artificial intelligence, represents a significant challenge to research integrity.
Elisabeth Bik, a microbiologist at Stanford University, helped expose the scale of the issue with a landmark 2016 study. Her paper, co-authored with Ferric Fang and Arturo Casadevall, analyzed over 20,000 published papers and documented hundreds of cases of image duplication and manipulation. Bik's work, for which she later won the 2024 Einstein Foundation Award, spurred a community of amateur scientific sleuths and served as a wake-up call for publishers.
Historically, scientific misconduct in physics was rare and often involved individual researchers. One high-profile case involved Bell Labs physicist Hendrik Schön, who was found in 2002 to have fabricated data in 16 papers. More recently, condensed-matter physicist Ranga Dias was found by a University of Rochester investigation to have committed misconduct, including data fabrication, regarding claims of high-temperature superconductors.
The Rise of Papermills
Today, misconduct has evolved into a coordinated commercial activity. So-called papermills are businesses that sell authorships, citations, data, and even complete manuscripts. They often market to students, promising guaranteed publication and citations, exploiting the academic pressure to "publish or perish." This pressure is linked to career advancement, grant applications, and, at some institutions, cash bonuses for publishing in high-impact journals.
The growth of open-access publishing has created new vulnerabilities. In this model, authors pay article-processing charges, which can range from around £2,500 to £10,000, instead of readers paying subscription fees. An analysis cited by the source found scientists paid over $1 billion in such fees to five major publishers between 2015 and 2018. Predatory journals exploit this system by collecting fees without conducting proper peer review.
The Scale of the Problem
The publication of fraudulent science is now outpacing the growth of legitimate publications, according to a 2025 study from Northwestern University. Data from the Retraction Watch Database, now part of Crossref, shows a sharp increase in retractions.
| Year | Retracted Papers |
|---|---|
| 2021 | 5,000 |
| 2022 | 6,000 |
| 2023 | 13,000 |
The database now contains over 60,000 retractions. Estimates suggest the number of papermill papers that entered the scientific record could be ten times higher.
New Tools and Tactics
Artificial intelligence and machine learning have given bad actors powerful new tools to generate fraudulent content and evade checks. In response, publishers are fighting back. IOP Publishing, which publishes Physics World, now employs a team of six research-integrity officers to handle misconduct cases, a problem described as far bigger than it was two decades ago.
Publishers have built up research integrity teams and developed new detection tools specifically to tackle papermill activity. The source notes that these efforts are a direct response to the significant rise in retracted articles in the early 2020s. Despite these countermeasures, fraudulent papermills remain a significant and evolving challenge within scholarly publishing.
