> - AIs becoming sentient and causing harm for their own ends.
You can scratch that one off the list. This is pure pseudo-intellectual fantasy. This notion isn't backed by anything and there are good reasons to disregard it, to put it mildly. It is fair to say that this concern isn't even wrong.
The concerns over "AI" have been and continue to be of a practical nature, namely, that of increasingly sophisticated automation. If we frame things in terms of automation, we can have a more productive discussion that stands of chance of contributing to good legislation and regulation instead of mystification and obscurantism.