DOI QR코드

DOI QR Code

Proposing CITRA-Based Design Ethics Guidelines for Deepfakes: A Comparative Analysis of the United States, European Union, and Japan

미국, 유럽, 일본 사례와의 비교를 통한 딥페이크 관련 CITRA 기반 디자인 윤리 가이드라인 제안

  • 조성배 (청주대학교 산업디자인학과)
  • Received : 2025.09.01
  • Accepted : 2025.09.18
  • Published : 2025.09.30

Abstract

This paper proposes CITRA (Consent & Dignity, Integrity, Traceability, Risk & Proportionality, Accountability & Remedy) design guidelines specialized for deepfakes, starting with the Ministry of Science and ICT's "Human-Centered AI Ethics Principles." By comparing and referencing norms, standards, and voluntary guidelines from the US, Europe, and Japan, and cross-analyzing consent, labeling/warnings, provenance, risk ratings, and accountability/remedy requirements based on the EU AI Act/DSA, NIST AI RMF and Generative AI Profile, Japanese government guidelines, C2PA (Content Credentials), and PAI recommendations, we reframe risks across the production-distribution-labeling stages along the CITRA axes and derive practical minimum compliance standards, including label visibility, persistence, and machine-readability; C2PA-based provenance; proportional controls in high-risk contexts; and service-level agreements (SLAs) for reporting, deletion, restoration, and support. This study presents key checkpoints and a blueprint for phased implementation that can be immediately applied in education, industry, and the public sector.

본 논문은 과학기술정보통신부의 '사람중심 AI 윤리 원칙'을 출발점으로, 딥페이크에 특화된 'CITRA; Consent & Dignity(동의 및 존엄성), Integrity(정직성/맥락보존), Traceability(추적 가능성), Risk & Proportionality(위험 및 비례성), Accountability & Remedy(책임 및 구제책)'디자인 가이드라인을 제안한다. 미국 유럽 일본의 규범·표준·자율 가이드를 비교 참조하고, EU의 AI Act/DSA, 미국의 NIST AI 위험관리 프레임워크 및 생성형 AI 프로파일, 일본 정부 지침, C2PA(Content Credentials), PAI 권고를 근거로 동의, 라벨링/경고, 출처성, 위험등급, 책임/구제 요구를 교차 분석하였다. 그 결과, 제작-유통-표시 단계의 위험을 CITRA 축으로 재구성하고, 라벨의 가시성·지속성·기계가독성, C2PA 기반 프로비넌스, 고위험 맥락의 비례 통제, 신고-삭제-복구-지원(SLA) 등 실천적 최소 준수선을 도출하였다. 본 연구는 교육·산업·공공 현장에서 즉시 적용 가능한 핵심 체크포인트와 단계적 도입 청사진을 제시한다.

Keywords

References

  1. 과학기술정보통신부, "사람이 중심이 되는 인공지능(AI) 윤리기준," 보도자료/정책자료, 2020년 12월
  2. Coalition for Content Provenance and Authenticity (C2PA), "C2PA Technical Specification, ver. 2.2". https://spec.c2pa.org/specifications/ (accessed Aug., 4, 2025).
  3. Partnership on AI, "Responsible Practices for Synthetic Media: A Framework for Collective Action", https://syntheticmedia.partnershiponai.org/ (accessed Aug., 5, 2025).
  4. 김봉섭, "2023년 사이버폭력 실태조사 결과와 의미", 지능정보윤리 이슈리포트, 제5권, 제1호, 23-34쪽, 2024년
  5. Sung Bae Jo, Jae Ick Lee, "Proposal of GUI Guidelines to Improve the Usability of Mobile Healthcare for New Silver Generation," Smart Media Journal, Vol. 7, No. 2, pp60-70, 2018 https://doi.org/10.30693/SMJ.
  6. National Institute of Standards and Technology (NIST), "Artificial Intelligence Risk Management Framework (AI RMF 1.0)", https://www.nist.gov/publications/artificial-intelligence-risk-management-framework-ai-rmf-10 (accessed Aug., 4, 2025).
  7. National Institute of Standards and Technology (NIST), "Artificial Intelligence Risk Management Framework: Generative AI Profile" https://doi.org/10.6028/NIST.AI.600-1 (accessed Aug., 5, 2025).
  8. European Commission, "Guidelines under the Digital Services Act to Mitigate Systemic Risks Online for Elections" https://ec.europa.eu/commission/presscorner/detail/en/ip_24_1707 (accessed Aug., 6, 2025).
  9. 정준화, "인공지능의 내재적 위험과 입법·정책 과제," 국회입법조사처 NARS 입법·정책, 제162호, 11-13쪽, 2024년 12월
  10. 강준모, "딥페이크 관련 국내외 규제 동향 분석," KISDI AI Outlook, 제18권, 8-13쪽, 2024년 9월
  11. European Commission, "Commission Publishes Guidelines under the DSA for the Mitigation of Systemic Risks Online for Elections," Press Release, Mar. 26, 2024.
  12. ByungRae Cha, MyeongSoo Choi, EunJu Kang, Sun Park, JongWon Kim, "Trends of SOC & SIEM Technology for Cybersecurity," Smart Media Journal, Vol. 6, No. 4, pp.41-49, 2017
  13. Digital Agency (Government of Japan), "The Guideline for Japanese Governments' Procurements and Use of Generative AI", https://www.digital.go.jp/en/news/3579c42d-b11c-4756-b66e-3d3e35175623 (accessed Aug. 9, 2025).
  14. Ministry of Economy, Trade and Industry (Japan), "AI Guidelines for Business Ver. 1.0", https://www.meti.go.jp/english/press/2024/0419_002.html (accessed Aug. 9, 2025).
  15. Dasol Kim, Sicheon You, "Concept and characteristics of safety information design that reflects human characteristics," Smart Media Journal, Vol. 13, No.8, pp.79-85, Aug. 2024 https://doi.org/10.30693/SMJ.