My money's on central intelligence (not the agency). With every service known to man publishing api's it's almost too easy to access information, and several services are now publishing api's to take actions (booking flights, hotels, etc.). I think we're much closer to a central intelligence learning to use these services on its own than a mobile intelligence that fits inside something of a reasonable size and of actual physical threat.

I don't see it as a new war. My bet is that it's going to be a continuation of the current war between legitimate software developers and organized crime running huge botnets. Today, the Storm Worm botnet is estimated to control 50 million computers and have more processing power than the top 500 supercomputers combined[1]. And, as computer security matures, organized crime will enhance their tools as well, and so on. It's a race, and both sides are using the latest technologies. And, eventually, it'll have to be AI against AI because humans won't be able to keep up.

[1] http://www.informationweek.com/news/201804528

A central "intelligence" called government (and other powerful behind the scenes groups) already manipulates the markets and the currency and all business outrageously. It already has significant say about what devices you do and do not have, what you can and cannot put into your body, what contracts you can and cannot form and enforce and takes half of your earnings. I think we need to consider the central intelligence that is already controlling so much.

In the short term that existing "intelligence" is likely to use AI, weak and eventually strong, to extend and deepen its control. I doubt very much that we will face the threat of a completely artificial general intelligence any time within the next couple of decades. Worrisome enough the trouble in our face today.

Hmm. I don't necessarily see a war in the short term between humans and some form of A.I. as inevitable. Even if an A.I. surpasses the common stock of humanity by the time that happens the technology and understanding of man will allow for self improvement at a vast rate. Probably it will be a moot point once it would be advantageous for a machine intelligence to go to war in the long term for there will be no point in a war since the lines will be blurry with biological components in computers and machine components in meat. Then any war will be over philosophical or economic grounds.