While I'm doing the translation, I read the whole story, and that's insane. I mean, what are they gonna do with those coordinates? Doesn't that mean a lot of people could potentially use the same service, and use it for something illegal too?
Well, I don't know the specifics of their scheme. But, from the research I've done, the idea is simple:
They create a set of images that give clues as to how the buyer can locate the product, which was previously hidden somewhere. In this way, the seller and the buyer never meet physically.
In the end, to an outsider, it just looks like another geo-coaching game.
The idea is for them to use platforms that host generic images, to make things less suspicious.
Because it's something so generic to an external observer, even an AI system would have difficulty detecting the problem in the initial uploads.
The images themselves don't show anything wrong; it's a photo of a place with some hand-drawn arrows. Automated systems would simply see this, and since they analyze each image individually, they won't find anything wrong.
So, the problem starts to become apparent when you look at the images together. Therefore, in these cases, only a human observer or a more advanced AI system could detect the issue. The analysis involves combining various images, the volume of images uploaded, and, in the case of TalkImg, the fact that they are not actually used here on the forum.
Typically, automated systems only detect images that explicitly violate the rules. Otherwise, only human verification or highly complex systems can detect them.
They probably try to do this with other image hosting services as well (it seems they did the same with Bitlist).
These weren't the first types of images that were sent that violated TalkImg's terms of use. But it was the most significant case because of its size.
