Comment by pizzly
9 days ago
If a computer vision model can identify ads and draw a rectangle over them, it can identify the DOM elements rendered inside that rectangle. Thus, the HTML, JS, and network calls of those ads can be detected. This data can then be used to train an ad detector model based on the DOM and network examples. The computer vision model essentially trains the DOM/network blocker, acting as a fallback mechanism when the blocker fails and continuously retraining the ad detector model.
Technically speaking, it’s possible to have the same kind of requests for legit content and ads. It’s also possible to randomize the response type (ad or content). In that case the only way to solve this before sending a request is to break their random number generator. That’s definitely not on the level of ad blockers.
However, I have no clue whether the content delivery system is already like that, and if not, then why. My domain knowledge is slim.
Not sure that's correct. You're replacing a simple visual recognition problem with something similar to the halting problem (but maybe less fatalistic, depending on the formalism you're working against ie DOM vs js).