Why Reverse Image Search Won't Find Your Stolen Art

Reverse image search (Google Images, TinEye) works by matching the file itself, or something very close to it: same pixels, same crop, same colours. That’s exactly why it’s so easy to defeat.

What a copycat actually does

Almost nobody re-uploads your image byte-for-byte. The pattern is more like:

Crop out your watermark or signature
Mirror the image horizontally
Shift the colour balance or apply a filter
Upscale it with an AI upscaler, which changes the underlying pixel data substantially
Re-compress it at a different quality level

Any one of these is usually enough to drop a reverse image search match entirely, because the tool is comparing pixel-level similarity, not the actual subject of the image.

What actually works

Content-based image matching, sometimes called perceptual hashing or embedding-based similarity, looks at what’s in the image rather than how the file is encoded. It’s the difference between asking “is this the same file” and “is this the same drawing, cropped and recoloured.” The second question is the one that actually matters when someone is trying to get away with it.

This is also why “just Google it” is bad advice to give someone who suspects their work has been stolen. It will confirm an exact re-upload exists. It says nothing about the far more common case: someone took your work and changed it just enough to dodge exactly that check.

What to do once you’ve found a match

Screenshot the infringing page in full, before contacting anyone. Pages disappear fast once their owner is notified.
Identify where it’s hosted. Our platform-by-platform DMCA guides cover the exact form and requirements for 18 platforms.
File with the host and request search delisting in parallel. Google’s removal process works independently of the platform-level report and often moves faster.

If you’d rather not manually re-check for reposts every few weeks, Fjerne does the content-based matching described above automatically and keeps watching after you file.