Perhaps we need OP to explain "I fully trust AI with these things", because it is difficult to define trust in this sense. Yes of course the human motivation was always going to be the real "AI threat", and fraud (and a hugely increased spam volume) was always going to be the main route of AI damage to society.