정보화정책 (Informatization Policy)

한국지능정보사회진흥원 (National Information Society Agency)
권호 표지
DOI QR코드

DOI QR Code

생성형 AI의 법적 문제와 규제 논의 동향

Legal Issues and Regulatory Discussions in Generative AI

  • 투고 : 2024.08.12
  • 심사 : 2024.09.04
  • 발행 : 2024.09.30
PDF 다운로드
⟨ 이전 논문 다음 논문 ⟩

초록

본 논문은 생성형 AI와 관련하여 제기되는 법적 문제점과 쟁점들을 정리한 것이다. 그리고 이러한 쟁점을 해결하거나 대응하기 위하여 혹은 생성형 AI가 제기하는 위험성을 최소화하기 위하여 개별 국가 또는 국제기구 등은 어떠한 규제적 논의들을 하고 있는지에 대하여 살펴보았다. 생성형 AI로 인해 제기되는 개인의 기본권 침해 문제, 새로운 범죄의 등장과 통제가능성, 특정 시장의 독점화 문제와 환경 문제 등이 주로 논의되고 있고, 규제의 필요성과 방향성에 대해 약간의 차이는 있지만 대부분의 국가들이 유사한 시각을 지니고 있는 것으로 보인다. AI와 관련하여서는 등장 초기부터 현재 제기되고 있는 문제들이 지속적으로 논의되어 왔었다. 특정 쟁점들은 상대적으로 많은 논의가 이루어졌지만 국가별로 조금씩 차이가 존재하기도 하고 과거와 다른 현상에 대한 고려가 필요한 상황들도 생겨나고 있다. 개별 국가의 상황에 맞추어 규제와 정책 등을 세밀화하고 있는 것으로 보인다. 다양한 쟁점들이 빠르게 등장하고 변화하는 상황에서, AI의 위험성을 최소화하고 안전한 AI의 활용을 통해 AI로 인한 효용과 이익을 향유하기 위한 방안들을 모색하여야 할 것이다. 국제적인 동향을 지속적으로 파악 및 분석하면서 국내에 적합한 AI 관련 규제와 세부 정책 등을 정비하는 것이 필요할 것이다.

This paper summarizes the legal problems and issues raised in relation to generative AI. In addition, we looked at what regulatory discussions individual countries or international organizations have in order to solve or respond to these issues or to minimize the risks posed by generative AI. Infringement of individual basic rights raised by generative AI, the emergence and control of new crimes, monopolization of specific markets and environmental issues are mainly discussed, and although there are some differences in the necessity and direction of regulation, most countries seem to have similar views. Regarding AI, the issues that are currently being raised have been discussed continuously from the beginning of its appearance. Although certain issues have been discussed relatively much, there are some differences between countries, and situations that require consideration of phenomena different from the past are emerging. It seems that regulations and policies are being refined according to the situation of individual countries. In a situation where various issues are rapidly emerging and changing, measures to minimize the risk of AI and to enjoy the utility and benefits of AI through the use of safe AI should be sought. It will be necessary to continuously identify and analyze international trends and reorganize AI-related regulations and detailed policies suitable for Korea.

키워드

참고문헌

  1. Abenezer G., Kidus M., Amit P., Anushka S., Vikas H., Vinay C. & Biplab S. (2024). "Privacy and Security Concerns in Generative AI: A Comprehensive Survey." IEEE Access Vol.12. 48126-48114.
  2. Ahn, S., Yoo J., Cho W., Noh J. & Son, H. (2023). "The Rise of the Supergiant Language Model and the Technical Characteristics and Social and Industrial Implications of ChatGPT." Issue Report IS-158, Gyeonggi: Software Policy & Research Institute.
  3. AiTIMES (2023). "Google succeeded in extracting personal information from 'Chat GPT'..."Can identify LLM training data." December 1.
  4. AiTIMES (2023). "GPT-4, 5 things that changed from before." March 15.
  5. AiTIMES (2024). "OpenAI unveils stage 5 to AGI..."We're on the verge of stage 2." July 12.
  6. Alexander, L., Martin, K. & Marc-Oliver P. (2024). "Blessing or curse? A survey on the Impact of Generative AI on Fake News." arXiv. https://arxiv.org/abs/2404.03021.
  7. Alexandra, S. L., Yacine, J. & Emma, S. (2024). "Power Hungry Processing: Watts Driving the Cost of AI Deployment?." FAccT '24: Proceedings of the 2024 ACM Conference on Fairness, Accountability, and Transparency, 85-99.
  8. Bianchi, F., Kalluri, P., Durmus, E., Ladhak F., Cheng, M., Nozza, D., Hashimoto, T., Jurafsky, D., Zou, J. & Caliskan, A. (2023). "Easily Accessible Text-to-Image Generation Amplifies Demographic Stereotypes at Large Scale." FAccT '23: Proceedings of the 2023 ACM Conference on Fairness, Accountability, and Transparency. 1493 - 1504.
  9. Bloomberg Law (2023). "Getty Images Sues Stability AI Over Art Generator IP Violations" February 7.
  10. Byun, S. (2023)."Ethical Problems of Super-Massive Generative AI." Journal of AI Humanities, 14, 63-82. https://doi.org/10.46397/JAIH.14.3
  11. Cabral, L., Haucap, J., Parker, G., Petropoulos, G., Valletti, T. & Van Alstyne, M. (2021). "The EU Digital Markets Act." JRC 122910. DOI: 10.2760/139337
  12. Carolina P. & Lorenzo P. (2024). "Artificial Intelligence and Cybersecurity." Intereconomics, 59(1), 10-13. https://doi.org/10.2478/ie-2024-0004
  13. Carugati, C. (2024). "The Generative AI challenges for competition authorities." Intereconomics, 59(1), 14-21. https://doi.org/10.2478/ie-2024-0005
  14. Chan, J. (2024). "How did OpenAI scrape the internet?" https://www.askhandle.com/blog/how-didopen-ai-scrape-the-internet. (Retrieved on August 2).
  15. CheckPoint (2023). "Cybercriminals Starting to Use ChatGPT."https://research.checkpoint.com/2023/opwnai-cybercriminals-starting-to-use-chatgpt/. (Retrieved on August 17).
  16. CNIL (2024). "AI How-to sheets(Ensuring the lawfulness of the data processing - Defining a legal basis)." https://www.cnil.fr/fr/node/164399. (Retrieved on August 5).
  17. Congressional Research Service (2023). "Generative Artificial Intelligence and Copyright Law." https://crsreports.congress.gov/product/pdf/LSB/LSB10922. (Retrieved on August 3).
  18. DeepMind (년도) "Tackling multiple tasks with a single visual language model." https://www.deepmind.com/blog/tackling-multiple-tasks-with-a-single-visual-language-model. (Retrieved on August 5).
  19. EDPB (2023). "EDPB resolves dispute on transfers by Meta and creates task force on Chat GPT." https://edpb.europa.eu/news/news/2023/edpb-resolves-dispute-transfers-meta-and-creates-task-force-chat-gpt_en. (Retrieved on August 15).
  20. EDPB (2024). "Report of the work undertaken by the ChatGPT Taskforce." https://www.edpb.europa.eu/our-work-tools/our-documents/other/report-work-undertaken-chatgpt-taskforce_en. (Retrieved on August 2).
  21. EDPS (2023). "G7 DPA Roundtable." https://www.edps.europa.eu/data-protection/our-work/ publications/international-conferences/2023-06-21-g7-dpa-roundtable_en. (Retrieved on August 3).
  22. European Commission IP Helpdesk (2023). "Intellectual Property in ChatGPT" Feburary 20.
  23. Federal Register (2023). "Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence." https:/www.federalregister.gov/documents/2023/03/16/2023-05321/copyright-registration-guidance-works-containing-material-generated-by-artificial-intelligence. (Retrieved on August 3).
  24. Feuerriegel Stefan, Jochen Hartmann, Christian Janiesch & Patrick Zschech. (2024). "Generative ai." Business & Information Systems Engineering 66(1), 111-126. https://doi.org/10.1007/s12599-023-00834-7
  25. Forbes (2023). "FTC Concerned About Generative AI Monopolies." June 30.
  26. Forbes (2023). "Six Risks Of Generative AI." June 29.
  27. FTC (2023). "Generative AI Raises Competition Concerns." https://www.ftc.gov/policy/advocacy-research/tech-at-ftc/2023/06/generative-ai-raises-competition-concerns. (Retrieved on August 7).
  28. G7 2023 Hiroshima Summit (2023). "Compendium of approaches to improving competition in digital markets." G7 2023 Hiroshima Summit.
  29. G7 Data Protection and Privacy Authorities Roundtable (2023). "Roundtable of G7 Data Protection and Privacy Authorities Statement on Generative AI." G7 Data Protection and Privacy Authorities Roundtable.
  30. Garante Per La Protezione Dei Dati Personali (2023). "Provvedimento del 30 marzo 2023 [9870832]: IL GARANTE PER LA PROTEZIONE DEI DATI PERSONALI." https://www.garanteprivacy.it/web/guest/home/docweb/-/docweb-display/docweb/9870832. (Retrieved on August 5).
  31. Goldman Sachs (2023). "Generative AI could raise global GDP by 7%." April 5.
  32. Google (2024). 2024 Environmental Report. Google. 
  33. Gupta, M., Akiri, C., Aryal, K., Parker, E. & Praharaj, L. (2023). "From ChatGPT to ThreatGPT: Impact of Generative AI in Cybersecurity and Privacy." IEEE Access, 11, 80218-80245. https://doi.org/10.1109/ACCESS.2023.3300381
  34. Hacker, P., Mittelstadt, B., Borgesius, F. Z. & Wachter, S. (2024). "Generative Discrimination: What Happens When Generative AI Exhibits Bias, and What Can Be Done About It." https://arxiv.org/abs/2407.10329.
  35. Han, J. (2023). "The Beginning of the Generative AI Era." Media Issue & Trend, 55. 6-17.
  36. Hong, E. (2023). "Beyond ChatGPT, Future of Generative AI (Generation AI) -Part 1."SAMSUNG SDS Insight Report. https://www.samsungsds.com/kr/insights/future_of_generative_ai_1.html, (Retrieved on August 9).
  37. IBM (2024). "What is a prompt injection attack?" https://www.ibm.com/kr-ko/topics/prompt-injection. (Retrieved on August 7.)
  38. ICO (2023). Joint statement on data scraping and the protection of privacy. ICO.
  39. ICO (2024). "Generative AI first call for evidence: The lawful basis for web scraping to train generative AI models." https://ico.org.uk/about-the-ico/what-we-do/our-work-on-artificial-intelligence/generative-ai-first-call-for-evidence/. (Retrieved on August 5).
  40. IEA (2024). Electricity 2024 Analysis and forcast to 2026. IEA
  41. Jeon, E. (2023). Personal Information Protection Issues Raised by Generative Artificial Intelligence Models such as ChatGPT. Regulatory Legislation Review, 23(3). Sejong: The Korea Institute of Legislation. 111-146.
  42. Jeong, C. (2023). "A Study on the Copyright Protection for Artificial Intelligence System." The Journal of Law, 31(2), 179-202. https://doi.org/10.35223/GNULAW.31.2.9
  43. Jung W. (2019). "Legal Issues and Policy Challenges on the Protection of Artificial Intelligence Creatures." Information and Communication Broadcasting Policy, 31(6), 1-27.
  44. Jungang Ilbo (2021). "[Factple] Learned human bias as it is, hate-spouting AI 'Iruda Shock'." January 12.
  45. Kim, B. (2022). "Legal Challenges in Large-Scale Language Models." Journal of Korea Infomation Law, 26(1), 173-217.
  46. Kim, H. (2023)."Could Data Lie?: Generative AI and Bias." IP & Data Law, 3(1), 75-105.
  47. Kim, J. (2023). "Generative AI, How far have we come (1) Basic concepts and development status." https://campaigns.do/discussions/598, (Retrieved on August 5).
  48. Kim, Y. (2016)."The Possibility to Protect Creative Works of ArtificialIntelligence." Chungnam Law Review, 27(3). 267-297.
  49. Kim, Y. (2023). "Legal Issues in Generative Artificial Intelligence Models: Focusing on the discussion of ChatGPT Products." Journal of Korea Infomation Law, 27(1), 77-112.
  50. Lee, S. (2023). "Issues and challenges in data law due to technological innovation." Journal of Law & Economic Regulation, 16(2), 84~106.
  51. Liang, W., Yuksekgonul, M., Mao, Y., Wu E. & Zou, J. (2023). "GPT detectors are biased against nonnative English writers." Patterns, 4(7). https://arxiv.org/abs/2304.02819.
  52. Mark, D. (2023). "Chat GPT and the EU AI Act: Will New Regulations be Added?" Denver Journal of International Law & Policy. https://djilp.org/chat-gpt-and-the-eu-ai-act-will-new-regulations-be-added/.(Retrieved on August 17).
  53. McAfee (2023). "Artificial Intelligence Voice Scams on the Rise with 1 in 4 Adults Impacted." https://www.mcafee.com/ko-kr/consumer-corporate/newsroom/press-releases/press-release.html?news_id=5aa0d525-c1ae-4b1e-a1b7-dc499410c5a1&PR=/PressMedia/09201999-C.asp&Sel=583. (Retrieved on August 3).
  54. MIT Technology Review (2020). "The way we train AI is fundamentally flawed."November 18.
  55. MIT Technology Review (2022). "We're getting a better idea of AI's true carbon footprint." November 14.
  56. Jovanovic, M. & Campbell, M. (2022). "Generative Artificial Intelligence: Trends and Prospects, Computer." Computer, 55(10). 107-112. https://doi.org/10.1109/MC.2022.3192720
  57. Nature (2024). "AI image generators often give racist and sexist results: can they be fixed?" https://www-nature-com-ssl.oca.korea.ac.kr/articles/d41586-024-00674-9. (Retrieved on August 5).
  58. NIST (2022). "Towards a Standard for Identifying and Managing Bias in Artificial Intelligence." (Retrieved on May 15).
  59. NIST (2024). "Adversarial Machine Learning A Taxonomy and Terminology of Attacks and Mitigations."NIST Trustworthy and Responsible AI NIST AI 100-2e2023.
  60. NIST (2024). "Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile." NIST Trustworthy and Responsible AI NIST AI 600-1.
  61. Novelli, C., Casolari, F., Hacker., Spedicato, G. & Floridi, L. (2024). "Generative AI in EU Law: Liability, Privacy, Intellectual Property, and Cybersecurity." arXiv. https://arxiv.org/abs/2401.07348
  62. Noy, S. & Zhang, W. (2023). "Experimental Evidence on the Productivity Effects of generative Artificial Intelligence." Science, 381(6654). 187-192. https://doi.org/10.1126/science.adh2586
  63. NVIDIA (2022). "What is a Transformer Model?(1)." https://blogs.nvidia.co.kr/2022/04/01/what-is-a-transformer-model/, (Retrieved on August 9).
  64. OECD (2007). "Glossary of statistical terms." https://stats.oecd.org/glossary/detail.asp?ID=3605. (Retrieved on August 7).
  65. Office of the Privacy Commissioner of Canada (2023). "OPC to investigate ChatGPT jointly with provincial privacy authorities." https://www.priv.gc.ca/en/opc-news/news-and-announcements/2023/an_230525-2/. (Retrieved on August 7).
  66. OneTrust DataGuidance (2024). "UK: Guidance on generative AI - the legal basis for scraping data." https://www.dataguidance.com/opinion/uk-guidance-generative-ai-legal-basis-scraping-data. (Retrieved on August 3).
  67. OpenAI. "DALL.E2." https://openai.com/product/dall-e-2. (Retrieved on August 5).
  68. OpenAI-written evidence (LLM0113) (2024). https://committees.parliament.uk/writtenevidence/126981/html/. (Retrieved on August 5).
  69. Park T. (2023). Park Tae-woong's lecture on AI. Seoul: Hanbit Biz
  70. Park, D. & Lee, H. (2024). "Literature Review of AI Hallucination Research Since the Advent of ChatGPT: Focusing on Papers from arXiv." Informatization Policy, 31(2), 3-38. https://doi.org/10.22693/NIAIP.2024.31.2.003
  71. Park, W., Kim, M., Park, Y., Ryu, H. & Jeong, D. (2024). "Taxonomy and Countermeasures for Generative Artificial Intelligence Crime Threats." Journal of the Korea Institute of Information Security & Cryptology, 34(2), 301-321.
  72. Personal Information Protection Commission (2023). Policy Direction for Safe Use of Personal Information in the Age of Artificial Intelligence. Seoul: Personal Information Protection Commission.
  73. Personal Information Protection Commission. (2024). Public Personal Information Processing Guide for Artificial Intelligence (AI) Development and Services. Seoul: Personal Information Protection Commission.
  74. Hacker, P., Engel, A. & Mauer, M. (2023). Regulating ChatGPT and other large generative AI models. Proceedings of the 2023 ACM Conference on Fairness, Accountability, and Transparency. DOI:10.48550/arXiv.2302.02337
  75. Reuters (2023). "Getty Images lawsuit says Stability AI misused photos to train AI." Feburary 7.
  76. Reuters (2024). "Microsoft, OpenAI plan $100 billion data-center project, media report says." May 30.
  77. Reuters (2024). "OpenAI's ChatGPT targeted in Austrian privacy complaint." April 29.
  78. Ryu, S. (2023). "A Study on Introducing a Specific Copyright Exception Clause to Text and Data Mining." Advanced Commercial Law Review, 101. 347-390.
  79. SBS News (2022). "[Exclusive] "AI Creation Can't Give You Copyright"...Domestic AI Copyright Conflict Ignition." October 14.
  80. Thome, S. (2024). "Understanding the Interplay between Trust, Reliability, and Human Factors in the Age of Generative AI." International Journal of Simulation: Systems, Science & Technology. 10.1-10.5.
  81. Son, S. (2016). "Copyright Protectionon Artificial Intelligence(AI) generated Works." Journal of Korea Infomation Law, 20(3), 83-110.
  82. Son, Y. (2023). "A Study on Creation by Generative AI and Copyright." Journal of Law and Politics research, 23(3). 357-389. https://doi.org/10.17926/KAOLP.2023.23.3.357
  83. Son, Y. (2024). "Generative artificial intelligence and personal information protection." Yonsei Law Review, 34(2). 351~382. https://doi.org/10.21717/ylr.34.2.11
  84. Song, M. & Lee, S. (2024). "What Concerns Does ChatGPT Raise for Us?: An Analysis Centered on CTM (Correlated Topic Modeling) of YouTube Video News Comments." Informatization Policy, 31(1), 3-31. https://doi.org/10.22693/NIAIP.2024.31.1.003
  85. Stanford University Human-Centered Artificial Intelligence (2023). "AI-Detectors Biased Against Non-Native English Writers." https://hai.stanford.edu/news/ai-detectors-biased-against-non-native-english-writers.(Retrieved on August 5).
  86. The NewYorkTimes (2023). "How Strangers Got My Email Address From ChatGPT's Model." December 22.
  87. The NewYorkTimes (2024a), "U.S. Clears Way for Antitrust Inquiries of Nvidia, Microsoft and OpenAI."June 5.
  88. The NewYorkTimes (2024b). "Regulators Take on the Giants of A.I." June 6.
  89. The White House (2023). "Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence." https://www.whitehouse.gov/briefing-room/presidential-actions/2023/10/30/executive-order-on-the-safe-secure-and-trustworthy-development-and-use-of-artificial-intelligence/. (Retrieved on August 8).
  90. The White House (2024). "A Call to Action to Combat Image-Based Sexual Abuse." https://www.whitehouse.gov/gpc/briefing-room/2024/05/23/a-call-to-action-to-combat-image-based-sexual-abuse/. (Retrieved on August 7).
  91. UK Parliament (2024). "Communications and Digital Committee Large language models and generative AI, 1st Report of Session 2023-24." https://publications.parliament.uk/pa/ld5804/ldselect/ldcomm/54/5402.htm. (Retrieved on August 5).
  92. UN WOMEN. (2023). "Glossary: Gender and Technology." https://www.unwomen.org/en/how-we-work/innovation-and-technology/glossary?gclid=EAIaIQobChMIjZe1p8zs_gIVWTBgCh3C1QAdEAAYASAAEgLYdvD_BwE. (Retrieved on August 7).
  93. UNESCO (2024a), "Challenging systematic prejudices: an investigation into bias against women and girls in large language models." https://unesdoc.unesco.org/ark:/48223/pf0000388971. (Retrieved on August 3).
  94. UNESCO (2024b). "Generative AI: UNESCO study reveals alarming evidence of regressive gender stereotypes." https://www.unesco.org/en/articles/generative-ai-unesco-study-reveals-alarming-evidence-regressive-gender-stereotypes. (Retrieved on August 3).
  95. Yang, J. & Yoon, S. (2023). "Going beyond ChatGPT to the era of Generative AI: Media and Content Generative AI Service Case and How to Secure Competitiveness." Media Issue & Trend Domestic Report, 55, 62-70.
  96. Yonhap News (2016). "Danger of Artificial Intelligence Brainwashing…MS Chatbot 'Tay' Suspended due to discriminatory remarks." March 25.
  97. Yoo, J., Ahn, S., Kim, J., Ahn, M., Jang, J., Bong, K. & Noh, J. (2023). "The Rise of Generative AI and the Change of Industry." ISSUE REPORT IS-161, Gyeonggi: Software Policy & Research Institute.
  98. 文化審議会著作権分科会法制度小委員会 (2024). "AIと著作権に関する考え方について." March 15.
  99. 文化庁 (2023). "AIと著作権". 令和5年度著作権セミナー. May 6.
  100. 朝日新聞デジタル (2023). "生成AI設計にプライバシー配慮を G7の個人データ保護当局が声明." June 21.
  101. 知的財産戦略本部 (2017). 知的財産推進計画2017.