The main reason I am not incredibly worried about bad AI outcomes is that every recent example turns out to be a combination of humans pushing an AI to do very specific bad things and marketing and straight up bs.

Will AI have the ability to be dangerous? Probably. But so far the greatest danger seems to be from businesses pushing their teams to do dangerous things, then loudly proclaiming in press releases about how their newest release is so dangerous and powerful and beyond human comprehension and powerful and powerful and powerful...