so is being a human being told that you're in the wrong by a robot that obviously has the wrong idea about the situation, but one side is paid to do it and is offered therapy and counseling.
aside : spare the human by using an AI for the first part; appeals need to be handled by humans. If the issue is the exposure to hatemail then all I can say is that humans in various industries have been able to cope with reading hatemail without major incident for some time now, as unpleasant as it may be.
Another aside : all employment is de-humanizing to some degree; the contract between worker and company is literally "let me pay you for the finite time you have on this earth so we can increase our financial bottom line." -- regardless of the mission statement it is rarely a motivation any more deep than that.
Let's at least keep the dehumanizing behavior focused on the people who are paid to experience it.