Shifting AI Governance in East Asia: AI Legislative Progress in Japan, South Korea and Taiwan

Shifting AI Governance in East Asia: AI Legislative Progress in Japan, South Korea and Taiwan

2025/09/09

Keywords: artificial intelligence, artificial intelligence regulation

I.Introduction

The landscape of AI governance in East Asia is changing, with two new AI laws enacted and one on the way. In South Korea, an act titled “the Basic Act on the Development of Artificial Intelligence and the Establishment of Foundation for Trustworthiness“ (“인공지능 발전과 신뢰 기반 조성 등에 관한 기본법”, henceforth referred to as “South Korea’s AI Act” or “SKAIA”)[1]was approved on December 26[2], 2024 and promulgated on January 21, 2025. The AI Basic Act is designed to establish a national AI governance framework and systematically foster the AI industry while preventing potential AI risks.[3] A few months later, Japan’s first law regulating AI was passed by the National Diet on May 28, 2025. The new law is titled "the Act on Promotion of Research and Development, and Utilization of AI-related Technology" (“人工知能関連技術の研究開発及び活用の推進に関する法律”, henceforth referred to as "Japan's AI Act" or "JAIA")[4], which reflects the strong will of the government to catch up in the global AI race.[5] Elsewhere in the region, Taiwan’s Executive Yuan finally passed its draft AI Basic Act (“人工智慧基本法草案”) on August 28[6] [7], which must now be submitted to the Legislative Yuan for deliberation. The government hopes the new law will lay the foundation for establishing Taiwan as an AI island and a key driving force in global AI development.[8] This article will give a quick overview of the key features of the three new AI regulations to illustrate the new landscape these countries are shaping in AI governance.

II.Key features of Japan’s AI Act (JAIA)

1.Purpose and principles of JAIA

Given Japan's lagging AI development and rising public concerns, JAIA reflects the government's worry about falling behind global peers in AI investment and adoption.[9] It is believed that new laws are needed in addition to existing laws and regulations to promote innovation and address risks.[10] Hence JAIA aims to advance the R&D and application of AI through the formulation of basic principles and plans, and the establishment of an "AI Strategic Headquarters".[11]

JAIA establishes basic principles for the promotion of the R&D and application of AI-related technologies[12], including enhancing industry R&D capabilities and competitiveness, systematically promoting AI collaboration from research to application with transparency, and enabling Japan to shape global norms through international cooperation.[13]

2.Industry Development and Promotion

JAIA requires the government to develop a National AI Basic Plan, in accordance with the basic principles, to promote the R&D and application of AI. The AI Basic Plan should set out fundamental policy guidelines and measures to comprehensively and systemically advance the R&D and application of AI-related technologies, along with other necessary provisions.[14]

JAIA also specifies basic measures to be included in the plan, which cover issues of promotion of R&D, expansion and sharing of facilities and data, human resources and education, international engagement in AI norm setting, and domestic guidelines making. In addition, the government should monitor AI technology trends and analyze cases of rights violations from improper AI use to develop countermeasures and provide guidance accordingly.[15]

3.Governance

JAIA stipulates that an AI Strategy Headquarters should be established under the Cabinet, composed of all cabinet members and headed by the Prime Minister.[16] The AI Strategic Headquarters is tasked with comprehensively and systematically advancing AI-related technology R&D and application policies, including the formulation, promotion, and implementation of AI Basic Plans and other related initiatives.[17] The Act also empowers the AI Strategy Headquarters to invite stakeholders to provide information, opinions or explanations, and other necessary assistance.[18]

4.Risk managements and rights protection

JAIA does not impose direct compliance obligations, but AI companies and research institutions are required to cooperate with government investigations and follow government guidance in cases involving violations of human rights and interests.[19]

5.Implementation of JAIA and Follow-up Work

JAIA came into force in May 2025. The Japanese government is required to develop guidelines that align with international standards and launch the Strategic Headquarters for the preparation and implementation of the National AI Basic Plan.

III.Key features of the South Korea’s AI Act (SKAIA)

1.Purpose and principles of SKAIA

SKAIA is designed to establish a foundation for AI development and trustworthiness, increasing citizens’ rights and interests protection, quality of life, and the country’s competitiveness.[20] It focuses on advancing national AI collaboration to foster a flourishing AI sector and developing legal frameworks to mitigate risks.[21]

Accordingly, the Act establishes basic AI development principles: prioritizing safety and reliability to improve quality of life, and ensuring those affected by AI output receive clear, meaningful explanations within reasonable parameters.[22]

2.Industry development and promotion

Supporting AI technology and industry development is a key feature of SKAIA. It establishes comprehensive measures covering technology development, industry revitalization, SME support, industrial foundations, talent cultivation, regulatory adaptation, and international cooperation.[23]

3.Governance

SKAIA also strengthens the institutional framework for AI governance. The Ministry of Science and ICT (henceforth referred to as “MSIT”) is mandated to execute an AI Master Plan every three years and empowered to investigate violations, require corrective action, and impose fines on non-compliant entities.[24]

The National AI Committee is authorized to review and decide on the AI Master Plan and AI-related matters, making it the highest decision-making body for South Korea's AI policies. It is composed of the heads of central administrative agencies and civilian AI experts appointed by the president.[25]

SKAIA also establishes the AI Policy Center to support MSIT on AI policy formulation, and the AI Safety Institute for AI safety matters.[26]

4.Risk management and rights protection

SKAIA imposes specific obligations on operators of high-impact AI and generative AI systems. All operators must ensure system transparency and safety, while high-impact AI operators face additional responsibilities including conducting fundamental rights impact assessments.[27]

High-impact AI systems are defined as AI systems that have a significant impact on or may pose a risk to human life, safety, and fundamental rights and are mainly utilized in critical infrastructure sectors and human rights-sensitive areas, or other areas specified by presidential decree.[28] The procedure for determining whether an AI system qualifies as high-impact AI will be established through subordinate legislation.[29]

5.Implementation of SKAIA and Follow-up Work

SKAIA will come into effect on January 1, 2026 and the formulation of subordinate statutes that detail enforcement mechanisms and guidelines should be expedited. However, domestic critics argue that corporate obligation provisions may hinder AI development and advocate for postponing their implementation.[30] Actually, an amendment to the Act was proposed in April 2025, seeking such a postponement along with a three-year grace period.[31]

IV. Key features of Taiwan’s draft AI Basic Act

1.Purpose and Principles of the draft AI Basic Act[32]

Taiwan adopts a relatively conservative approach to AI policy and measures to boost industrial development have long occupied the agenda of AI governance. Given that AI is a crucial technology for national development, the draft AI Basic Act (henceforth referred to as "the draft Act") seeks to ensure that AI technology develops vigorously in a human-centered approach, encourage innovation while considering human rights, and safeguard Taiwan’s national sovereignty and cultural values.[33]

Hence, the draft Act establishes seven guiding principles in line with international norms, which are sustainability, human autonomy, privacy protection and data governance, security, transparency and explainability, fairness and accountability.[34]

2.Industry Development and Promotion

It is the government’s responsibility to promote the R&D and application of AI and construct the infrastructure needed.[35] In order to facilitate AI innovations, competent authorities may provide a controlled environment for testing and validating AI innovation products and services before they are released to the market or put into use.[36] Considering the wide scope of AI application and development, the government is encouraged to collaborate with the private sector, including through public-private partnerships, and should promote international cooperation on AI matters.[37] The government should also continue to comprehensively promote AI education at all levels to enhance the public's AI literacy.[38]

Data is crucial for AI development, so the draft Act mandates the government to establish mechanisms to enhance data availability, and measures to facilitate AI outputs that maintain the country's multicultural values, and protect intellectual property rights.[39]

3.Risk Management and Rights Protection

(1) Risk Management

The draft Act includes several provisions addressing AI risks. The government should take steps to prevent AI from being used for illegal purposes. For example, Ministry of Digital Affairs (MODA) and other relevant agencies may provide or recommend tools or methods for AI evaluation and verification to avoid misuse of AI.[40] Secondly, MODA is mandated to foster an AI risk classification framework, based on which sectoral competent authorities should establish risk-based tiered management standards.[41] Thirdly, the government may, through binding regulations or non-binding administrative guidance, promote safety standards, verification, transparent and explainable traceability, or accountability mechanisms to enhance the trustworthiness of AI development and application.[42] Lastly, the government should clarify the ownership and conditions of liability for high-risk AI applications and establish relevant mechanisms for relief, compensation or insurance to protect affected parties.[43] However, AI application responsibility norms would not apply to pre-release activities in order to support technological innovation.[44] [45]

(2) Rights Protection

The draft Act concerns not only the privacy rights of individuals but also labor rights. The government should ensure the protection of personal data used throughout the AI lifecycle on the one hand[46] , and also protect workers' rights and provide necessary assistance to help them adapt to technological changes, especially those who have lost their jobs due to AI use.[47]

4.Governance and Implementation

Despite the heated debate regarding the designation of a dedicated AI regulatory authority in the country, the Executive Yuan decided against establishing such an authority, given AI's cross-ministerial nature. Relevant competent authorities will be responsible for formulating implementing regulations and guidelines and the Executive Yuan will continue to guide relevant agencies and departments at all levels through the existing Digital Legal Coordination Meeting to facilitate the development of AI.[48]

V.Analysis and conclusion

Japan, South Korea and Taiwan all seek to maintain the countries' momentum in promoting AI development through AI legislation. The three parties all emphasize trustworthy AI, though they actually place greater emphasis on AI development. They share considerable common ground in the policies to foster AI industry development, such as promoting AI R&D and application and supporting infrastructure-building, and diverge in their approaches to addressing potential AI-related risks and governance structure.

Japan adopts a ‘light touch’ regulatory approach to AI regulation, maintaining coherent policy coordination that responds to domestic imperatives and global trends without imposing regulatory burdens on industries.[49] The country favors a soft approach with governmental guidance. In contrast, South Korea incorporates regulatory provisions specifically targeting high-impact AI systems in its AI Basic Act, seeking to balance between enhancing national competitiveness through AI and mitigating potential risks stemming from AI misuse, though this approach actually faces some domestic opposition currently. Taiwan adopts an approach similar to Japan's. The draft AI Basic Act avoids imposing regulatory obligations, and the government will prioritize AI verification and evaluation mechanisms to ensure trustworthy AI development.

Regarding governance approaches, both Japan and South Korea seek to strengthen governmental AI governance functions through legislation, with Japan establishing an AI Strategic Headquarters and South Korea creating an AI Committee, both operating under their respective Cabinets. In contrast, Taiwan's draft AI Basic Act does not address governance structural matters.

Given the profound societal transformations that AI technology may bring, all three East Asian countries recognize the importance of sustained AI advancement while acknowledging the critical need to ensure AI safety and trustworthiness to protect human rights. In an era of intense global AI competition, it seems to be the best policy for governments to carefully design AI policies that strike a balance between fostering innovation and safeguarding human rights. This cautious approach is essential as significant challenges remain and AI risks demand comprehensive solutions.

Reference:

[1] 인공지능 발전과 신뢰 기반 조성 등에 관한 기본법（법률 제20676호, 2025. 1. 21, 제정），법제처 국가법령정보센터，https://www.law.go.kr/%EB%B2%95%EB%A0%B9/%EC%9D%B8%EA%B3%B5%EC%A7%80%EB%8A%A5%20%EB%B0%
9C%EC%A0%84%EA%B3%BC%20%EC%8B%A0%EB%A2%B0%20%EA%B8%B0%EB%B0%98%20%EC%A1%B0%EC%
84%B1%20%EB%93%B1%EC%97%90%20%EA%B4%80%ED%95%9C%20%EA%B8%B0%EB%B3%B8%EB%B2%95/(206
76,20250121) （最後瀏覽日：2025/09/11）。

[2] A New Chapter in the Age of AI: Basic Act on AI Passed at the National Assembly‘s Plenary Session, Ministry of Science and ICT, https://www.msit.go.kr/eng/bbs/view.do?sCode=eng&mId=4&mPid=2&pageIndex=&bbsSeqNo=42&nttSeqNo=1071&searchOpt=ALL&searchTxt= (last visited Sept. 11, 2025).

[3] A New Chapter in the Age of AI: Basic Act on AI Passed at the National Assembly‘s Plenary Session, Ministry of Science and ICT, https://www.msit.go.kr/eng/bbs/view.do?sCode=eng&mId=4&mPid=2&pageIndex=&bbsSeqNo=42&nttSeqNo=1071&searchOpt=ALL&searchTxt= (last visited Sept. 11, 2025).

[4] 人工知能関連技術の研究開発及び活用の推進に関する法律（令和7年法律第53号），e-Gov法令検索，https://laws.e-gov.go.jp/law/507AC0000000053（最後瀏覽日：2025/09/11）。

[5] CABINET OFFICE, GOVERNMENT OF JAPAN, Outline of the Act on Promotion of Research and Development, and Utilization of AI-related Technology (AI Act), https://www8.cao.go.jp/cstp/ai/ai_hou_gaiyou_en.pdf (last visited Sept. 11, 2025).

[6] 〈政院通過「人工智慧基本法」草案 建構AI發展與應用良善環境 打造臺灣成為AI人工智慧島〉，行政院，https://www.ey.gov.tw/Page/9277F759E41CCD91/5d673d1e-f418-47dc-ab35-a06600f77f07（最後瀏覽日：2025/09/09）。

[7] There are other AI bills brought up by legislators in the Legislative Yuan. The purpose of this article is to analyze the AI governance priorities of the governments of Japan, South Korea, and Taiwan; therefore, other AI bills proposed by legislators are not included in the discussion.

[8] 蘇文彬，〈行政院通過AI基本法草案，將不設立AI專責機關〉，iThome，2025/08/28，https://www.ithome.com.tw/news/170874 (最後瀏覽日：2025/09/09)。

[9] Japan’s AI Bill Advances Toward Enactment, Connect on Tech (May 27, 2025), https://connectontech.bakermckenzie.com/japans-ai-bill-advances-toward-enactment/ (last visited Sept. 9, 2025).

[10] 松尾剛行，〈【2025年施行】AI新法とは？AIの研究開発・利活用を推進する法律を分かりやすく解説！〉，Keiyaku-Watch，https://keiyaku-watch.jp/media/hourei/2025-ai-law/（最後瀏覽日：2025/09/11）。

[11] 人工知能関連技術の研究開発及び活用の推進に関する法律（令和7年法律第53号）第1条。

[12] 人工知能関連技術の研究開発及び活用の推進に関する法律（令和7年法律第53号）第3条。

[13] Japan Enacts AI Promotion Act: Overview and Implications for Businesses, Zelo Law Square (May, 2025), https://zelojapan.com/en/lawsquare/56899 (last visited Sept. 9, 2025).

[14] 人工知能関連技術の研究開発及び活用の推進に関する法律（令和7年法律第53号）第18条。

[15] 人工知能関連技術の研究開発及び活用の推進に関する法律（令和7年法律第53号）第11-17条。

[16] 人工知能関連技術の研究開発及び活用の推進に関する法律（令和7年法律第53号）第19、21-24条。

[17] 人工知能関連技術の研究開発及び活用の推進に関する法律（令和7年法律第53号）第20条。

[18] 人工知能関連技術の研究開発及び活用の推進に関する法律（令和7年法律第53号）第25条。

[19] 人工知能関連技術の研究開発及び活用の推進に関する法律（令和7年法律第53号）第16条。

[20] 인공지능 발전과 신뢰 기반 조성 등에 관한 기본법，제1조。

[21] The Korean AI Basic Act: Asia’s First Comprehensive Framework on AI, Lexology (Mar. 17, 2025), https://www.lexology.com/library/detail.aspx?g=f91ff0fb-94ed-4aa9-b667-65d6206a7227 (last visited Sept. 9, 2025).

[22] 인공지능 발전과 신뢰 기반 조성 등에 관한 기본법，제3조。

[23] 인공지능 발전과 신뢰 기반 조성 등에 관한 기본법，제13-26조。

[24] 인공지능 발전과 신뢰 기반 조성 등에 관한 기본법，제40조。

[25] 인공지능 발전과 신뢰 기반 조성 등에 관한 기본법，제7조。

[26] 인공지능 발전과 신뢰 기반 조성 등에 관한 기본법，제6-12조。

[27] 인공지능 발전과 신뢰 기반 조성 등에 관한 기본법，제31-32조。

[28] 인공지능 발전과 신뢰 기반 조성 등에 관한 기본법，제4조。

[29] 인공지능 발전과 신뢰 기반 조성 등에 관한 기본법，제33조。

[30] Seungmin (Helen) Lee, South Korea’s Evolving AI Regulations, Stimson (June 12, 2025), https://www.stimson.org/2025/south-koreas-evolving-ai-regulations/ (last visited Sept. 9, 2025).

[31] 〈인공지능 발전과 신뢰 기반 조성 등에 관한 기본법 일부개정법률안〉，대한민국국회，https://likms.assembly.go.kr/bill/bi/billDetailPage.do?billId=PRC_N2M5K0S3R2R0Q1O3X5X1W1U1T7P3Q6&currMenuNo=2600044（最後瀏覽日：2025/09/09）。

[32] 〈政院通過「人工智慧基本法」草案 建構AI發展與應用良善環境 打造臺灣成為AI人工智慧島〉，行政院，https://www.ey.gov.tw/Page/9277F759E41CCD91/5d673d1e-f418-47dc-ab35-a06600f77f07（最後瀏覽日：2025/09/09）。

[33] 人工智慧基本法草案第1條。

[34] 人工智慧基本法草案第3條。

[35] 人工智慧基本法草案第4條。

[36] 人工智慧基本法草案第5條。

[37] 人工智慧基本法草案第6條。

[38] 人工智慧基本法草案第7條。

[39] 人工智慧基本法草案第14條。

[40] 人工智慧基本法草案第8條。

[41] 人工智慧基本法草案第9條。

[42] 人工智慧基本法草案第10條。

[43] 人工智慧基本法草案第11條。

[44] 人工智慧基本法草案第11條。

[45] See also: Taiwan Rolls Out Draft Artificial Intelligence Law, OCACNEWS, July 18, 2024, https://ocacnews.net/article/374412 (last visited Sept. 3, 2025).

[46] 人工智慧基本法草案第14條。

[47] 人工智慧基本法草案第12條。

[48] 蘇文彬，〈行政院通過AI基本法草案，將不設立AI專責機關〉，iThome，2025/08/28，https://www.ithome.com.tw/news/170874 (最後瀏覽日：2025/09/09)。

[49] Sun Ryung Park, Less Regulation, More Innovation in Japan’s AI Governance, East Asia Forum (May 21, 2025), https://eastasiaforum.org/2025/05/21/less-regulation-more-innovation-in-japans-ai-governance/ (last visited July 4, 2025).