And the automatic answer is: 'But it can be risky, depending on what it's used for' or if this occurs might it be than asked by it: What will I use humans for.

Another words more than risky the 'super-machine' might and probably would declare itself boss automatically. You could try to program a fail-safe kill switch in this machine but I would not bet on it working when you really need it.

Seriously humans should already be continuously monitoring the
internet and other obvious and not so obvious places for the possibility of a complete take over of so-called humanity before humanity knows what has happened.

This theoretical super-machine may be far more subtle and much
more dangerous than any reptellian, gray or other alien life form imagined to date and it may not be friendly.

Further it can be postulated that humans are already, by building faster and more efficient artificial intelligence,
working for a super-brain - and humans have already lost what is sometimes called 'free-will'.

-Give me more ram or else!