Also, remember that the power to disable machines still remains with humans, until machines become totally independent of humans
I disagree here, as if you imagine that an AI on the internet now influences decision making, a general intelligence AI would be able to at a highly granular level across the globe. It wouldn't even need to attack, just manipulate. When it comes to humans will, they will do what they have always done, use their tools to maximize - but this tool could come to think for itself.
You seems sure for something that is supposed to be a hypothesis.
I am confident, not sure. When it comes to the future technology - it is uncertain what will arrive - when it comes to human behavior, we have a lot of evidence we can use to predict.