설명가능한 생성형 AI 결과물과 영업비밀보호법 쟁점

Explainable Generative AI, which aims to increase the reliability of generative AI, shows the reasoning process that generated the result when generating and displaying the result requested in the prompt. Compared to existing versions that only display the output, explainable generative AI offers various advantages, such as allowing users to easily verify the reasoning process. However, the explanation of the reasoning process that generated the output may contain content that violates personal information or copyrights. In particular, if it displays information related to intellectual property rights such as patents, it may infringe on trade secrets. Output generated by generative AI cannot be registered as an inventor under intellectual property rights such as patents. Therefore, it is difficult to find ways to protect trade secrets in the following cases. First, when a product or service invented by a leading company using generative AI was commercialized and generating revenue, later entrants purchased the leading company's product or service, inputted images into the generative AI, or used explainable generative AI to acquire the core technology of the product and obtained the same results. Since the core technology was not acquired through illicit means, there was no infringement of trade secrets. Therefore, this paper examines the issues that may arise from the perspective of the Trade Secret Protection Act when the results of explainable generative AI are used for commercial gain and proposes solutions. We propose the establishment of a "specialized organization to manage products, works, and services utilizing generative AI results." This specialized organization would form a "committee" of field experts to evaluate the extent to which generative AI results have been utilized in registered works, products, and services. The committee would assess the value and grant "generative AI property rights."