The Legal Nature of Artificial Intelligence Inventions

Abstract

Artificial intelligence, due to its lack of independent legal personality and the absence of attribution of civil liability, faces gaps regarding AI related inventions. This study examines the legal nature of inventions based on AI technologies through a comparative approach between Iranian and Afghan law. With the expansion of machine learning algorithms and generative content systems, the definition and protection of AI inventions have been challenged within the framework of national laws. The main objective is to identify legal models compatible with the specific characteristics of such inventions (including reproducibility, explainability, and the role of humans in the innovation process). This study investigates two main axes: first, whether the output generated by AI models can be considered an invention or merely the result of human machine collaboration; and second, how the role of a specific human actor can be defined in such cases. The research methodology relies on comparative studies of international and domestic laws, examination of key judicial cases, and a combined legal technical analysis. The findings indicate that despite technological advancements, most legal systems still attribute AI inventions to “human produced outcomes” or “human controlled tools,” and patent applications are typically filed in the name of the AI user, the person who enabled the AI operation, or companies possessing technical expertise. It is recommended that new legal frameworks or specific amendments be developed to address these technological configurations.