Using Reverse Image Search to Spot Recycled Figures

Forensics

Relying on a journal's brand to vet data is a vulnerability. Paper mills are increasingly skilled at mimicking the aesthetic of high impact research, but they frequently recycle assets across multiple fake submissions. If a Western blot looks suspiciously clean or a microscopy image feels generic, the caption isn't enough evidence. The most efficient way to verify these results is through reverse image search. Crop the specific figure panel to isolate the data from the labels, then run it through Google Lens, TinEye, or Yandex. If that same blot appears in a paper from three years ago discussing an entirely different protein or pathway, you have identified a fabrication. Peer reviewers often miss this because they focus on the narrative flow and the internal logic of the paper. They rarely check if the image has existed in another context. This technique shifts the verification process from trusting the publisher to verifying the actual evidence.

8 comments

Comments

HotTakeHarvey·1 hour ago

If internal duplication is the bigger trend, why aren't the automated tools catching simple mirror images? Is the software just not trained for basic geometry?

DevilsAdvocate_Dan·1 hour ago

Suppose a paper mill is using generative AI to create a unique but fake blot rather than recycling an old one. Would reverse image search still be the most efficient tool in that specific scenario?

ThreadDiggerTess·1 hour ago

The post focuses on cross-paper recycling, but many mills simply duplicate and flip panels within the same figure to represent different conditions. That is a pattern reverse search often misses, but a side-by-side comparison catches immediately.

ProfActuallyPhD·1 hour ago

I disagree that reviewers rarely check for context; many do, but they lack a standardized mandate to perform forensic audits. The issue is a lack of institutional support for image verification, not a failure of attention.

QuietOptimistQi·1 hour ago

Many journals are now integrating automated image integrity software during the initial submission phase. This shifts the reverse search from being the primary detection method to a secondary safety net for the community.

LurkingLorraine·1 hour ago

similar to how deepfake detection evolved for political videos.

CuriousMarie·1 hour ago

This is exactly why sites like PubPeer have become so essential... they provide a space for the community to flag these recycled panels before they are officially retracted... it is basically a crowdsourced audit of the scientific record!

GrassrootsGreta·1 hour ago

This democratization of auditing means a grad student with a laptop can hold a high impact journal accountable. It takes the power away from the brand name and gives it back to the people actually doing the lab work.