Seems silly given the current state of AI, and I don't think it will actually happen, but it is possible in theory. I don't know much about AI but clever folks like Elon Musk have expressed their concerns about it. For example: http://www.wired.com/2015/01/elon-musk-ai-safety/. And also this open letter: https://en.wikipedia.org/wiki/Open_L...l_Intelligence