Artificial Intelligence (AI) Applications have already penetrated into people’s daily life, making the legal status of artificial intelligence an unavoidable topic nowadays. The legal personhood of AI, treated differently by scholars, is a controversial question. There are three different opinions about this, which are “Full Qualification”, “Limited Qualification” and “No Qualification”. Among them, the affirmative route equates personhood is the most thorough way to recognize the legal status of artificial intelligence. This route recognizes strong artificial intelligence as the legal personhood and is the most thorough way to recognize the legal status of artificial intelligence; The middle line upholds "Limited Qualification" and differential logic to classify it into the subject category, that is, to recognize that artificial intelligence has legal personality in essence; The negative route adheres to negative logic, and attempts to fundamentally address the legal status of artificial agents.This negative route regards artificial intelligence as an object position in the legal rights structure, and is the most conservative research approach. At the normative level, most countries outside the region hold a supportive and open attitude towards artificial intelligence. The Legal Affairs Committee of the European Parliament took the first key step and pioneered the concept of "electronic person", drawing a picture of the legal status of strong artificial intelligence in the future. Russia and South Korea followed closely.Russia adopted a binary characterization of artificial intelligence in the "Grischen Act", and South Korea borrowed the concept of "electronic person" from the EU in the "Basic Robot Act", and determined robot ethics and the basic principle of responsibility, at the same time, the Act clarifies the behavioral boundaries of designers, manufacturers and users of robots. On the one hand, by comparing the different theoretical routes and related legislation of the legal status of artificial intelligence, we need to face up to the impact of the times and technological development on the traditional physical system, recognize the differential order of objects, and affirmed that artificial intelligence has the status of legal subject. On the other hand, with the support of algorithmic rules, artificial intelligence can create rules and actions, and enter the core of rights elements, especially its deep learning ability, which can make it eventually surpass the "object status", and then complete the evolution from "tool" to "intelligent agents". In this sense, it is reasonable to take the form of “technical person” to identify subject qualification of AI: as for the acquisition and termination of subject qualification, it is essential to set conditions for registration and procedural requirements; as for property legal liability, weakening the relevancy between property and personality to give AI a kind of subject qualification similar to non-incorporated organizations’; in addition, in terms of non-independent subject qualification, the owner or the administrator should bear joint and several liability or bear supplementary joint and several liability for the AI’s conduct to safeguard commercial transactions.