Any AI in the foreseeable future will be under control of human beings, either due to laws or financial ownership. I'm not the least worried about AI, but having watched this election, the humans in my country scare the shit out of me.
I agree that it seems implausible that an AI will "rebel" and set its own goals, that is still sci-fi. But ordinarily those scary men would have to enlist the help of many others, like the old DDR (East Germany) where almost a hundred thousand of a population of 16 million were STASI agents. Use a NSL, use a computer to transcribe it and analyze who is talking to who when and about what and for how long and you could have a quite Orwellian database with only a hundred people involved. Same thing with bank transactions, electronic tickets, number plate scanners, facial recognition, deep packet scanning and whatnot the automatic collection and processing enables a very small but organized and powerful fraction of the population to surveillance and control the rest.
I'm not really sure we could reverse that trend because just like we might discuss the pros and cons of everyone having tiny little digital cameras on them at all times it is still extremely unlikely to change. The average person is leaving more and more electronic traces and even those who try to avoid it is often found by metadata from their friends or they stand out by being the black hole that isn't sharing. Very often the system is rigged towards it, for example we didn't want bus drivers to get robbed so much so now it's a lot more expensive to pay cash than a prepaid subscription or cell phone payment, both of which link to a real identity here in Norway. People do what's cheapest so the few people who pay cash stand out. And really it's no problem until it is a problem, but then you're usually too late to change it.