How do paper mills reuse the same lab images across batches of fake studies without getting caught by detectors?

This explores how paper mills reuse images across batches of fabricated manuscripts while avoiding detection. The direct answer is that this collection does not cover it. None of the retrieved notes study paper mills, duplicated scientific figures, or image-forensics tools, so the corpus cannot tell you which tricks mills use. Common examples are cropping, rotating, recoloring and splicing one source image into several papers. Treat everything below as a related pattern from other fields, not as evidence about paper mills.

The closest idea in the corpus is about where a check is applied, not about images. In the work on malicious agent skills, attackers beat six different scanners because each scanner judged one skill at a time. The attackers used the scanners' own feedback to make each piece look harmless, while the harmful plan only appeared when the pieces were combined Can attackers evade skill scanners by refining individual skills?. A paper mill would face the same weakness if journals screened each submission on its own instead of comparing it with other submissions. That is an analogy for you to test against the image-integrity literature, not something these notes show.

A second related idea is that disguising a source and fooling a detector are separate claims. A note on heavily rewritten text points out that a paper claimed its rewrites also beat AI-text detectors but never ran a detector test Do rewrites that hide authorship also fool AI detectors?. The same caution applies to claims about altered images: hiding where an image came from does not prove that a particular detector will miss it.

The defensive notes have a similar shape. One method spots planted documents in a retrieval system by masking parts of each document and flagging the ones that behave abnormally Can we defend RAG systems from corpus poisoning without retraining?. Another shows that probing repeatedly can tell decoys from genuine objects when the two respond differently, though only in an idealized setting Can repeated quiet probes separate decoys from genuine objects?. Both suggest the same lesson: look for oddities across a whole collection instead of judging items one by one. To understand paper mills themselves, you will need sources outside this library, such as research-integrity and image-forensics work.