The rapid development of intelligent technology and its application has a strong impact on the contemporary legal system based on the fundamental premise of human subjectivity, and the criminal law status of artificial intelligent subjects has become a key issue. At present, the negative theory and the positive theory are in a stalemate, accelerating the process of knowledge change in the current criminal law theoretical system. Different degrees of negativism, such as "artificial objects", have their merits on the grounds that the current criminal legal system is difficult to directly accept them and they carry potential criminal risks. Its logical mechanism is the presupposition effect of excessive release of anthropocentrism. Examining from an anthropocentric standpoint will lead to irreconcilable conclusions, and it also shows the tendency of the tool attributes of intelligent technology applications to be deeply magnified. However, the above-mentioned negative reasons are not necessarily in line with the actual needs of regulating artificial intelligence crimes, they are also inconsistent with the evolution law of legal subject system, and it will even suppress the autonomous evolution of the subject of criminal law. Different forms of positive theory such as "cyborg" are the first to break out of absolute anthropocentrism and its conceptual constraints, and view the interaction and trends of artificial intelligence and criminal law from a developmental perspective, which is more in line with the development trend of artificial intelligence crimes. The accumulation of existing knowledge frameworks and new elements, such as freedom of will, criminal responsibility, moral and ethical rules, intelligence level and types of intelligent subjects, the objective existence of criminal responsibility, and the special status and significance of algorithms, means that artificial intelligence can develop in the future to achieve cohesion and fit in the criminal law system. This reversal of understanding, which does not really completely deviate from the principles of criminal law, also provides support at the theoretical level. Moreover, following the rationale of utilitarianism, and under the guidance of a series of advanced legislation on artificial intelligence subjects, the criminal status of intelligent subjects should be determined in stages, types, and dynamics, so that they can continue and expand on the path of legal fiction. Under the clarification of the subject status of artificial intelligence in criminal law, it is necessary to accept the reality that intelligent subjects can enjoy certain emerging rights, but the specific content such as the type of rights should be specifically confirmed. Under the instigation of the current technology generation, the content and scope of the rights of intelligent subjects are currently limited, and it is impossible to adopt a protection strategy equivalent to that of "people", and the protection methods should also be different. This kind of "dimension reduction" protection has its practical rationality. Through active system design and rule configuration, it can coordinate with the changing situation of the criminal law status of artificial intelligence subjects to the greatest extend.