It's not just cartels, but also self-citations. For example, the original preprint of this paper [1] has 5 self-citations, with the version accepted to EMNLP increasing that to 10. This kind of behavior is sadly pretty common now and when you mention it to conference organizers they just shrug.