Hey Maciek!
Great question!
The assets I’ve shipped so far have all been audio based, so for that type of asset I was imagining something similar to how Shazam is able to create a fingerprint for audio. I think it generates hashes at different intervals of time through the source and then potential candidates are found and scored based on the total number of matched hashes.
My initial idea was to maybe generate a hash from the data of each file, a bit like a checksum. That should be an easy find for if something is directly copied and re-uploaded, however if a scammer was to make minor changes to all the files, then this method would fall flat. It could be a good basic test to catch out lazy scammers.
I don’t think it’ll be an easy but it would be reassuring to know that there’s some protection in place