Tracking the retracted paper mill articles: a bibliometric study

Abstract Paper mills have been a growing challenge to the integrity of academic research, yet there has been a need for more empirical research on the impact, spread, and scope of the problem. This bibliometric study uses the Retraction Watch database to identify all of the paper mill papers that have been retracted. The search yielded 10,409 retracted paper mill papers since the database’s inception until 2024, with a massive spike in 2023. While research articles were most common, the results also showed conference abstract/papers, spiking in 2022, had gone through paper mills. The results show paper mill papers could be found from around the world, but the highest concentration came from China by far (9,137), representing 87.8% of all identified paper mill papers. The study identified over 97,093 co-author pairs from across the world in these papers. Papers labeled as Business and Technology, Business and Life Sciences, and Health Sciences were the most prevalent in the findings, including critical fields such as computer science and cancer research. The most common connected keywords from 2011–2016 were primarily biomedical, including cells , cell , and cancer . This switched towards computer science from 2021–2024, with common keywords such as model , data , and algorithm. Relatively few paper mill papers were found in the social sciences or humanities. Given the growth of the threat to the integrity of science, especially within the medical fields and with added AI availability, the results join the chorus of concerns calling for more research on paper mills across academic publishing.