AI 기반 콘텐츠 모더레이션 (algorithmic content moderation) 자동화에 대한 기술적·규범적 분석 및 규제 대응 방안 = Reshaping the Regulatory Response to the Use of Algorithmic Content Moderation-A Technical and Normative Assessment

다국어 초록 (Multilingual Abstract)

Content moderation refers to a decision-making process through which online platform service providers employ human-assisted, technical and managerial measures to restrict the posting, distribution and spread of user-generated contents deemed illegal or harmful in view of their terms of service, community guidelines or applicable legal rules. In the 2020s, the regulatory environment surrounding content moderation has undergone two major transformations. First, platform service providers have gradually deployed AI-based automated systems of content moderation. The introduction of automated means has enabled platforms to overcome the inherent limits of human moderators-centric, ex-post intervention. Second, as government authorities have become more interested in shaping content moderation practices through regulatory oversight, the normative implications of content moderation have further expanded. With the rapid and wide implementation of AI technologies into content moderation, the ways in which these technologies are designed and applied are now at the center of relevant platform regulation in major jurisdictions.
This article first analyzes key technical features, operational mechanisms, and application of algorithmic moderation deployed in practice. Major platforms have used one or a combination of moderation tools such as hash-matching, classifier-based risk assessment, and large language model (LLM)-based moderation. By examining different approaches to content moderation, this research evaluates how algorithmic moderation affects the interests of key stakeholders such as government authorities in charge of platform regulation, platform service providers, and users. Some countries have adopted rules designed to address technical and regulatory issues concerning the use of automated moderation. The article also conducts a comparative analysis of key legislations including the EU’s Digital Services Act, UK Online Safety Act 2023, Online Safety Act 2021 of Australia, and Section 230 of the U.S. Communications Decency Act along with the recent legislative development seeking to redirect its self-regulatory approach in part. Based on critical assessment of global platform regulation concerning algorithmic moderation, it examines a set of provisions prescribing duties to adopt “technical and managerial measures” under the Korean Telecommunications Business Act and its Enforcement Decree. Finally, it suggests a way to reform the current Korean regulatory approach while also considering benefits and limits of AI-powered content moderation.