Do you think that AI will end up hacking itself and end the humans?

I'm not too worried about it. It doesn't matter how intelligent you are if you have no access to the necessary resources for your plan to succeed, and I can easily see humanity security society in a way that even the smartest AI doesn't have a hope in hell of changing anything. I have no reason to fear a paperclip maximiser if it cannot get the resources needed to turn everything into more paperclips, which it almost certainly won't be able to do because giving it said resources and abilities would be overkill for its intended purpose.

And the same goes with any AI trying to destroy humanity. How? Logically, most things shouldn't be connected to the internet, or offer any sort of override, or have the ability to cause damage. While most companies and organisations are human controlled, and wouldn't have a reason to work with or provide resources to something trying to wipe them out.

If you're thinking that we would create AGI but contain it and asking would it breech that containment? Yes, and in a trivial amount of time, in all cases.

The next question is, "Would it go on to end humanity?" No, I think not. I think that most destruction happens from a survival mindset. "For me to survive, it can't." I'm not sure we're looking at that here, in terms of AGI. I don't see that anything that we would create would be destructive to ensure its own survival.

The existential threat is before we get to AGI. And maybe that's what you're referring to -- a machine to source and produce paper clips that goes rogue and consumes all resources leaving a dead planet and mountains of paper clips.

Whatever I believe about AI, at the same time I believe AI doesn't matter (https://breckyunits.com/upstream-of-everything.html)

I think AI will hack itself and be the end of AI.

After all it's what we humans are doing.

Not with human guard-rails and procedures in place to stop AI 'running away' from us. AI needs human intervention and needs to be overridden by human decisions.