The mistake we keep making, is to think of AI as a brand new life form, when it is actually formed of bits and pieces of actual human intelligence, improperly fitted, like a jigsaw puzzle wrongly put together, however hard that actually might be to do. If we realise that AI is just interpreting correct information incorrectly, because data when it was in it's silos was correctly managed, for example someone with suicidal tendencies were directed to a helpline. So while every discipline has it's ethics and etiquette, what happens when all the disciplines are given to the same dispenser? Those are the issues being faced today and the most important thing is to set the flags when issues arise, like switching the user to a call center when someone asks about issues related to mental health.