The study aimed to theoretically determine the normative challenges arising from the emergence of artificial superintelligence (ASI) capable of autonomous thinking and decision-making beyond human epistemic control. The interdisciplinary analysis considered the universality of the human rights concept, the normative boundaries of moral subjectivity, and the possibility of granting AI a legal status. The study presented a conceptual typology of legal models of superintelligence status, including instrumental, limited-subjective and full-fledged paradigms, with an indication of their advantages and risks. In particular, the instrumental model retains full human control but loses its normative relevance in the context of system autonomy; the limited subjectivity model can be used to delegate responsibility within certain limits without violating the principle of human supremacy; the full-fledged model, which equates ASI with a legal entity, questions the current ethical framework of legal personality. The main results of the study demonstrated that the anthropocentric legal doctrine is insufficient to consider the cognitive multiplicity of agents who do not have bodily vulnerability but demonstrate a high level of autonomy, reflexivity and adaptability. The study established that the cognitive asymmetry between a human and a superintelligent agent generates a new form of epistemic injustice that makes it impossible to participate equally in the procedure of moral decision-making. The study proposed the concept of limited legal personality as a normative compromise which ensures legal certainty and delimitation of liability between the participants of interaction. The results have implications for the philosophy of law, regulatory policy in the field of AI, and interstate regulatory regulation. They can be used to form international approaches to the certification of autonomous systems, guarantee the explainability of algorithmic decisions and preserve human normative autonomy in the era of cognitive multiplicity