Knowing that it is possible for a representation to be falsified (what you describe the ad network as knowing) is very far from knowing that it is false (the branch of the standard you appear to be appealing to argue a lack of right-to-rely.) So, I don't think the argument, as you've made it, is convincing.