Agreed. This problem is closer to "online graffiti" (every text box is the canvas) than sexism, racism, etc. The assumption that a bot may be on the receiving end further encourages the behavior. These kids know better, though I do think a healthy "fear of consequences" or reality check can help.

I'd expect the rate of abuse to decline fairly significantly with the simple addition of the text "This conversation is being monitored by a supervisor", appearing a few moments after the first message is sent.