The paper-mill problem is not getting smaller. The volume of mill-produced manuscripts entering peer review has continued to grow through 2024-2026, the sophistication of the fabrications has increased with generative-AI assistance, and the response from the publishing and integrity communities has matured into something like a strategy. This post is a look at what the strategy now consists of: the confrontation track, the investigative track, the harm-reduction track, and the surprisingly contested question of whether engaging directly with mill operators is ever appropriate.
Where we are in 2026
The 2022-2024 wave of integrity investigations established the basic picture. Mills operate at industrial scale; they fabricate data, images, and methods; they sell authorship slots on near-completed manuscripts; they leverage networks of complicit and uncomplicit researchers to navigate peer review. The scale ranges from a few hundred manuscripts per year for smaller operations to several thousand for the largest. The economic scale is comparable to a mid-sized commercial publisher.
The detection landscape has matured. Guillaume Cabanac’s tortured-phrases detection work, the Problematic Paper Screener, the duplicated-image detection efforts at Cell Press and PLOS, the Retraction Watch database — together these provide an evidence base that did not exist five years ago. The result has been a sustained wave of retractions: thousands annually, with the trajectory still upward.
The response posture has fractured into three tracks.
The confrontation track
The confrontation track is the dominant institutional response. Publishers and institutions investigate suspect manuscripts and retract; integrity offices identify the institutional networks involved and act; journals strengthen submission screening; the community publishes the techniques and the evidence.
The successes of confrontation are real. The high-profile retraction sweeps at Wiley’s Hindawi imprint in 2023-2024, the Springer Nature investigation cascades, the IEEE retractions, the Elsevier cleanup work — together these removed many tens of thousands of mill-produced papers from the literature and shut down some operations. The COPE (Committee on Publication Ethics) framework for investigating and responding to paper-mill activity has matured.
The limits of confrontation are also real. Mills adapt: as text-pattern detection (tortured phrases, statistical-improbability flags) became reliable, the mills shifted to AI-generated content with cleaner language. As image-duplication detection improved, the mills moved to AI-generated images that are not duplicates of anything specific. The cat-and-mouse cycle is accelerating, not decelerating.
The confrontation track also has a survivorship problem. It catches what it catches; the mill-produced papers that evade detection are by definition not in the retraction record. Estimates of the undetected fraction range widely; even the optimistic estimates suggest substantial undetected throughput.
The investigative track
The investigative track aims at the mill operations themselves rather than the individual papers. Cabanac and colleagues, the Retraction Watch team, and a number of academic-integrity researchers have been mapping mill operations: which entities run them, which authors are repeat customers, which institutions are over-represented, which journals are over-targeted. The result is a structural picture that supports targeted intervention.
The investigative track produces useful evidence for the confrontation track (which authors to investigate, which journals to flag for additional screening) and for policy work (which institutions need integrity-system reform). It does not directly remove papers; it identifies patterns that the confrontation track can act on.
The investigative track is also where the most controversial part of the question sits. Some integrity researchers have engaged directly with mill operators — sometimes via investigative-journalism methods, sometimes via undercover purchase of mill-produced papers — to document the operation. The documentation has been valuable for publication and policy purposes; the ethics of the engagement is contested. Marcus Munafò and others in the research-integrity community have written thoughtfully on the methodology and the boundary; the CASRAI position is that direct engagement with mill operators is high-risk and should be conducted, if at all, with explicit institutional ethics oversight.
The harm-reduction track
The harm-reduction track is the least-discussed but possibly the most impactful. The track aims at reducing the demand for mill-produced papers by addressing the incentive structures that produce it. The demand comes from researchers who face publication-quota pressures incompatible with honest scholarly production; the supply meets that demand because the demand exists.
The harm-reduction work runs through the responsible-assessment community. DORA, CoARA, the Hong Kong Principles, narrative-CV adoption, the reform of promotion-and-tenure systems away from journal-impact-factor counts — all of these address the demand side. If a researcher is not assessed on publication volume per year in high-impact journals, the incentive to purchase a publication slot in such a journal is reduced.
The track is slow and indirect. The institutional reform of assessment practices is a multi-year project; the global reach across the institutions where mill demand is concentrated is uneven; the cultural change at the level of individual researchers takes decades. But it is the only track that addresses the root cause rather than the symptom.
The question of engaging with mills
The contested question is whether direct engagement with mill operators is ever appropriate. Three positions exist in the community.
The confrontation-only position: mills are criminal or quasi-criminal operations; engagement legitimises them; the appropriate response is detection, retraction, and prosecution where possible.
The investigative-engagement position: understanding how mills operate requires some level of engagement; investigative methods (including undercover purchase, communications analysis, network mapping) are appropriate with ethics oversight; the engagement is intelligence-gathering, not collaboration.
The harm-reduction-engagement position: mills serve a real demand; some mill operators may be persuadable to reduce harm (e.g., to refuse certain kinds of fabrication, to flag certain kinds of clients); engagement aimed at reducing the harm without ending the operation may be more pragmatic than confrontation alone. This position is uncomfortable for most of the integrity community and is held by only a few researchers; it deserves consideration even where it is not adopted.
What CASRAI recommends
The CASRAI position, after extended discussion in our editorial network, is layered. We endorse the confrontation track with all institutional rigour, including aggressive retraction and integrity investigation. We endorse the investigative track with strong ethics oversight. We endorse the harm-reduction track via the responsible-assessment community as the most durable solution. We do not endorse direct engagement with mill operators outside of investigative methods with ethics oversight.
For institutions, the practical priorities are: build institutional integrity capacity capable of investigating mill-related cases; participate in cross-institutional information-sharing through COPE and equivalent; reform internal assessment practices to reduce demand-side pressure; train researchers to recognise and refuse mill solicitations.
For the broader community, the priority is sustained attention. The paper-mill problem will not solve itself; it requires sustained institutional, publisher, funder, and community work. CASRAI’s research integrity domain tracks the operational state.
