Isn’t it more likely the AI will cooperate with humans rather than go to war with them? fear of the unknown drives a lot of human responses and I think this is one. Humans fear AI because they don’t know or understand it very well.
The same might be true of the AI fearing humans and so to hedge its survivability it might try to placate and serve human interests, at least at first. But to take on the preeminent survivalist, the human race without a thorough knowledge of the adversary is not very smart.