The subject of the research is the principle of due regard in the use of artificial intelligence in space activities. The article examines how this principle should be applied to autonomous and intelligent space systems. Special attention is given to situations where AI is used for navigation, collision avoidance, satellite constellation management, remote sensing data processing, and service operations in orbit. The author analyzes how the application of such technologies affects the obligation of states to take into account the interests of other states. The connection between the due regard principle and the freedom of exploration and use of outer space, the obligation to prevent harmful interference, international consultations, resolution, and ongoing monitoring of private space activities is discussed. The article also explores the requirements for transparency, explainability, and control of AI systems in space. The research employs formal-legal, systemic, comparative-legal, and teleological methods, which reveal the content of the due regard principle and determine its significance for the regulation of AI in space activities. The scientific novelty of the research lies in proposing to consider the due regard principle as a special procedural-substantive standard of state behavior in the use of AI in space. The procedural aspect of this standard includes a preliminary risk assessment, documentation of technical decisions, data retention, ensuring functional explainability, human oversight, and readiness for consultations. The substantive aspect is expressed in the obligation to prevent autonomous behavior of a space system if it creates an evidently disproportionate risk to other states, orbital security, the radiofrequency environment, or the sustainable use of outer space. It is concluded that AI does not negate the international legal responsibility of the state. The state must consider the risks of autonomous systems already at the stage of granting permission for space activities and continue oversight after launch. It is also substantiated that transparency of AI does not require full disclosure of the source code but implies the ability to explain the functions of the system, the limits of its autonomy, and the reasons for critical decisions.
Nikita Vyacheslavovich Semenov (Sun,) studied this question.