I believe it is theoretically possible for AI to become as intelligent as humans. This shouldn't be a great cause for concern though. Everything that AI can do is programmed by humans. Perhaps the question could be phrased differently: "Could robots be dangerous?" Of course the could be! If humans programs robots to destroy and do bad things, then the robots could be dangers. That's basically what military drones do. They are remotely controlled, but they are still robots.