It has emerged that "paper mills" that create fake papers for money are operating widely around the world. Ads are flooding the internet claiming to manipulate not only authorship but also citation counts, and evidence has surfaced that fake papers have actually been published in major journals, badly damaging trust in science.
A joint team from Germany and the United States said on the 23rd (local time) on Zenodo that after examining about 19,000 ads posted online over six years by seven paper mills, they suspect fake papers are being widely distributed in the scientific community.
The researchers examined 18,719 ads posted online from March 2020 through early this month. The median price for first authorship sold by paper mills was about $800 (about 1.18 million won), and depending on a paper's value or a journal's prominence, transactions reached more than $5,600 (about 8.27 million won).
◇ Temptation of academic fraud spreads worldwide
The study was conducted by Anna Abalkina of Free University of Berlin and Reese Richardson and Spencer Hong of Northwestern University. Abalkina exposed paper mills that churn out fake papers and was named one of Nature's "10 people who led science in 2024." The other two are postdoctoral researchers in Luís Amaral's lab.
The team collected 2,311 ads posted on the Telegram messenger app by three ghostwriting firms in India, Iraq and Uzbekistan. They also identified 16,399 ads on the websites of four firms believed to operate in Russia, Latvia, Kazakhstan and Ukraine.
All the firms advertised authorship and even offered services such as listing as a textbook author, patents, copyright registration and fake awards. Richardson said, "The term 'paper mill' alone can't capture everything that's happening now, given the range of businesses they run," adding, "I see them as corporations operating in a 'reputation manipulation market.'"
Fake papers are suspected of being published in leading international journals. Nature's news team found that as many as 53 papers whose titles matched ads from three paper mills were actually published. Twenty-three appeared in journals published by the Institute of Electrical and Electronics Engineers (IEEE), the world's largest engineering society, and five and four were published in journals from Wiley and Springer Nature, respectively.
But only five of these papers have been retracted. One was pulled at the in-press stage. The rest remain in academic databases and are being cited by researchers. Nature's news team said it notified six publishers that carried fake papers. The publishers were said to have begun investigations after being informed.
◇ Performance pressure drives fake papers
Behind the boom in paper mills lies academia's rigid incentive structure that evaluates research performance solely by the number of papers and citations. Abalkina wrote in Nature in June last year, "The growth in the number of researchers and the requirement to publish in international journals introduced by many countries are two reasons paper mills emerged." In fact, a Saudi Arabia assistant professor told Nature that despite publishing 20 papers in a single year, additional performance pressure led the person to pay hundreds of dollars to a paper mill to add their name to a fake paper.
This phenomenon is particularly severe in medicine. Gengyan Tang of the University of Calgary studied Chinese teaching hospitals and found that overburdened nurses and young doctors facing research-performance demands are major customers of paper mills. According to Nature's review of retractions from 2014 to 2024, seven of the 10 institutions with the highest retraction rates were hospitals in China.
Keeping pace with academic demand, paper mills are rapidly evolving. According to Abalkina's analysis, a Russia-based paper mill generated about $6.5 million (about 9.6 billion won) in sales by selling co-authorship between 2019 and 2021. Abalkina estimated the industry's scale at hundreds of millions of dollars a year.
Output from paper mills is also surging. According to a paper released last year in the Proceedings of the National Academy of Sciences (PNAS) by Amaral's team at Northwestern, the number of papers suspected to be produced by paper mills is doubling every 1.5 years. That pace far outstrips retractions, which double every 3.3 years, showing that current fraud-detection technology and cleanup systems are failing to keep up with paper mills' expansion.
◇ AI emerges as a game changer in paper fraud
Advances in artificial intelligence (AI) are expected to worsen paper fraud. In April last year, Oxford University hosted a three-day academic event discussing paper mills. Experts predicted that Generative AI, which can instantly produce hard-to-detect fake data, images and text, will be a "game changer" that reshapes paper fraud.
Scientific publishers are developing in-house tools to filter out fake papers and sharing related information in response. But building systems to determine whether a paper came from a paper mill is costly. Abalkina noted, "Publishers gain financially from publishing papers, so they have little interest in changing their systems or slowing publication speed."
Science's value lies in integrity. Because the spread of paper mills that buy and sell reputation can ultimately topple the edifice of human knowledge, fundamental measures are needed to reduce market demand. Richardson of Northwestern told Nature, "We need to go beyond filtering individual papers and address the root causes of why researchers turn to such services."
References
Zenodo (2026), DOI: https://doi.org/10.5281/zenodo.19684278
Nature (2025), DOI: https://doi.org/10.1038/d41586-025-01824-3
PNAS (2025), DOI: https://doi.org/10.1073/pnas.2420092122