머신러닝 알고리즘의 데이터 처리에 대한 법적 제한의 한계 :개인정보보호와 차별금지의 측면에서

This article addresses the issues of how the evolution of machine learning field has an impact on the legal system. In the area of automated data processing, a legal model for data privacy protection is applied in a manner that regulates the collection and use of personal data. This model assumes primarily with the personal identification or identifiability. However, if we look closely at the process of collecting and using these data, we can confirm that the legal model for anti-discrimination based on the principle of anti-classification can also be applied in an overlapping manner. I analyze several legal cases in order to identify the overlapping points. In each cases, there are some legal issues about automated data processing for data privacy and/or anti-discrimination. Data processing performed by machine learning algorithm which is trained with dataset, is divided into three steps: collection(or selection), analysis, and application. Based on these distinctions, limitations of legal regulations for data processing that can be encountered in each step are reviewed through models for data privacy protection and anti-discrimination.