Although progress has been made in the current security defense theories in cyberspace and various types of security technologies have come into being, the effectiveness of such theories and technologies often depends on the scale of the prior knowledge of the attackers, on the part of the defender and on the acquired real-timing and accuracy regarding the attackers’ behavior features and other information. Hence, there lacks an efficient active defense means to deal with uncertain security threats from the unknown. Even if the bottom-line defense technologies such as encrypted verification are adopted, the security of hardware/software products cannot be quantitatively designed, verified or measured. Due to the “loose coupling” relationship and border defense modes between the defender and the protected target, there exist insurmountable theoretical and technological challenges in the protection of the defender and the target against the utilization of internal vulnerabilities or backdoors, as well as in dealing with attack scenarios based on backdoor-activated collaboration from both inside and outside, no matter how augmented or accumulated protective measures are adopted. Therefore, it is urgent to jump out of the stereotyped thinking based on conventional defense theories and technologies, find new theories and methods to effectively reduce the utilization of vulnerabilities and backdoors of the targets without relying on the priori knowledge and feature information, and to develop new technological means to offset uncertain threats based on unknown vulnerabilities and backdoors from an innovative perspective.