AI 발명의 특허법적 한계와 가능성 - 발명자 인정 기준과 진보성 심사 기준의 재구성 -

초록

This study aimed to analyze the core legal issues that ‘AI inventions’—those where human inventive contribution exists, but AI is utilized as a tool for exploration and analysis—face within the current patent system of the Republic of Korea, and to propose institutional measures for improvement. A legal gap currently exists in effectively accommodating and protecting the results of such inventions because the current Patent Act restricts the definition of ‘inventor’ to a natural person (human being). The research conducted an in-depth examination of three critical issues related to the patentability of AI inventions. First, regarding the inventorship requirement, which is directly tied to the subject of patent rights, an analysis of the constitutional premise that regards ‘human creation’ as the essence of invention and major international precedents, such as the ‘DABUS case,’ confirmed the global trend of not recognizing AI as an inventor. It was thus proposed that AI be considered an advanced auxiliary tool, limiting inventorship to humans, but that there is an urgent need to establish specific checklists and guidelines—similar to the USPTO's ‘significant contribution’ standard—to assess the substantive creative intervention by the human inventor. Second, concerning the non-obviousness (inventive step) standard, as AI's capabilities challenge the traditional ‘Person Having Ordinary Skill in the Art (PHOSITA)’ concept, the study proposed new examination criteria to address the difficulty in determining inventiveness. This criterion emphasizes that while results derived merely from large-scale data Volume processing should be restricted from obtaining non-obviousness recognition, the standard should pivot to key assessment factors such as the human inventor's creative efforts in the Quality of data selection and processing, and the critical modification and integration of the AI-generated output to achieve a conspicuous technical effect. Third, regarding the issues of ownership and liability, the necessity of an AI Ethics and Governance Framework was asserted to prevent ‘false inventorship’ and to clearly define the responsible party should social harm arise from the invented product. In conclusion, this research logically proposes that, to legally support technological innovation in the AI era and secure the legitimacy of the patent system, the Korean Intellectual Property Office (KIPO) must enhance predictability and transparency by clarifying the criteria for determining the human's substantive contribution to AI inventions. Furthermore, it suggests comprehensive system improvements, such as mid-to-long-term amendments to the Patent Act to clarify the concept of the inventor, the attribution of rights, and the locus of liability.

키워드