The article addresses the problem of determining the civil law status of artificial intelligence systems. The study analyses the four principal approaches developed in international legal doctrine on this issue: recognition of artificial intelligence as an object of civil law, the electronic agent model, the concept of limited legal personality (quasi-subject), and the idea of electronic personhood. The article reveals the normative origins of the electronic agent construction, its attributive function in contractual relations, and its limitations in the field of tort law; the quasi-subject model is assessed on the basis of the theory of divisible legal capacity and the criteria of legitimacy. It is further demonstrated that the concept of electronic personhood cannot be justified under the interest criterion, since it does not give rise to any new interest requiring protection but, on the contrary, creates the risk of establishing a liability shield. As a result of the study, the article examines the applicability of the institutions of representation, transactions, and torts under the Civil Code of the Republic of Uzbekistan to relations involving autonomous systems, and puts forward proposals for introducing a special regime for artificial intelligence while retaining its status as an object of law.