The expansive connectivity of information systems has set the stage for pervasive malware to leverage multiple attack vectors and propagation methods. In doing so, this malware has taken on the complexity and richness of the very society it endeavors to control. Defending against it is therefore exceptionally difficult because defense systems have no autonomy in perceiving threats of complex malware and reacting against it. In this paper, smart malware model is defined as emerging complex malware that may be used by defense systems to perceive complex malware and reacting to its attacks. A targeted attack is also presented to show the difficulty of defending systems against smart malware. It is also compared with conventional malware to analyze malware types. Moreover, a numerical study about smart malware is presented to evaluate the proposed model in a more precise manner. The comparison and the numerical study show that our model can be used to perceive smart malware autonomously by automated tools. Copyright (C) 2017 John Wiley & Sons, Ltd.