The "danger" of their policies (and I've benefitted from them, too) is that they obviously can be gamed, and they obviously have to have defenses against that - which means if you cross some invisible line (and now likely AI-monitored) you're doomed; no recourse.