AI 학습용 데이터의 보호에 관한 소고

A Study on the Protection of AI Training Data - Focusing on Protection under the Intellectual Property Act

데이터는 AI 기술과 밀접한 관계를 가진다. 데이터가 축적되면 될수록, 데이터가 정확하면 할수록 좋은 분석결과가 나온다. 이러한 분석에는 AI 기술(AI 소프트웨어)이 활용된다. 딥러닝과 머신러닝과 같은 AI 기술의 발전은 데이터를 더욱 가치 있게 만들어 왔다. 데이터는 AI 기술과 함께 거래되는 경우가 많아지게 되었고, 데이터와 AI 기술을 활용한 비즈니스도 더욱 활기를 띠고 있다. 이러한 관점에서 AI 학습용 데이터는 적절히 보호될 필요가 있다.

Data is closely related to AI technology. The more data is accumulated and the more accurate the data is, the better the analysis results come out. AI technology (AI software) is used for this analysis. Advances in AI technologies such as deep learning and machine learning have made data more valuable. Data is often traded with AI technology, and businesses using data and AI technology are also becoming more vibrant. From this point of view, AI learning data needs to be adequately protected. In the case of the domestic data industry, data creation and utilization is evaluated as relatively inadequate. Data required for data construction and utilization (distribution) is insufficient, and industrial and social use is poor due to a closed distribution system. This phenomenon is believed to be due to the limited use of data due to restrictions on personal information, and the lack of manpower responding to corporate demand. In particular, learning data essential for AI-related inventions is no exception. In this study, the following measures were proposed to protect AI learning data. First, a plan to strengthen protection under the patent law, second, a plan to protect data through the Unfair Competition Prevention Act, third, a plan to introduce a data patent application system in preparation for the era of big data, and fourth, Fourth, similar to the microbial donation system, the introduction of the AI system for depositing data and learning completion models was suggested.