[ADRN Issue Briefing] South Korea’s Approach to AI Governance for Democracy

Introduction

The disruptive impacts of AI on democracy range from deepfakes and cognitive outsourcing to electoral manipulation in democratic systems, as well as panopticon-like surveillance in authoritarian regimes. Ovadya argues that dystopias of autocratic centralization and ungovernable decentralization can be avoided through “democratic centralization and democratic decentralization,” enabled by four innovations: mini-public deliberation, AI-augmented decision-making, third-party participatory governance, and platform democracy.[1]

While this approach—seeking to advance both AI innovation and democratic governance—is reasonable, much of the current debate on AI’s impact on democracy focuses primarily on procedural dimensions, often overlooking AI’s potential to link citizen demands to public services. Significant room remains for AI-assisted government innovation to improve service delivery. If implemented wisely, AI-assisted public governance could help narrow service gaps in an era of widening wealth inequality. This briefing examines how South Korea is responding to public demands for both substantive democracy and legitimate procedural democracy.

South Korea’s Recent AI Policy Initiatives

South Korea passed the Framework Act on the Development of Artificial Intelligence and the Establishment of Trust—commonly referred to as the AI Basic Law. The law was adopted by the National Assembly on December 26, 2024, formally promulgated on January 21, 2025, and entered into force on January 22, 2026.

The AI Basic Law establishes a comprehensive regulatory framework that both promotes the development of the AI industry and strengthens safety, ethical standards, and user protections, particularly for high-impact AI systems. It provides for a Presidential Committee to serve as a national AI “control tower,” a dedicated AI Safety Institute, and requires the government to formulate an AI Master Plan every three years to enhance national competitiveness. Under the Act, AI-generated content must be clearly labeled—including through watermarks—and developers and operators of “high-impact” AI systems must implement appropriate safety measures, alongside industry development measures such as R&D funding, training data construction, and the designation of AI clusters.

The law’s regulatory scope extends to foreign AI systems affecting users or markets within South Korea. Foreign AI companies without a domestic presence must designate a local agent if they meet certain thresholds—such as having more than one million average daily users in Korea or generating substantial domestic revenue. The Ministry of Science and ICT (MSIT) is empowered to investigate violations and may issue corrective orders or impose administrative fines of up to KRW 30 million (approximately USD 22,000).

This marks a shift from voluntary guidelines to mandatory compliance with AI ethics. The previous “National Guidelines for Artificial Intelligence (AI) Ethics,” adopted in December 2020 by the Presidential Committee on the Fourth Industrial Revolution, rested on three basic principles: respect for human dignity, the common good of society, and the proper use of technology.[2] These guidelines relied on a safeguard and self-regulation mechanism, deliberately avoiding rigid legal norms deemed inadequate for a fast-changing technological environment. Concerns about excessive regulation hampering industrial growth were also a factor. With the AI Basic Law, however, South Korea became the first country to enforce binding regulations on high-impact AI, while the European Union’s AI Act continues to delay its full implementation.

Since President Lee Jae-myung took office in May 2025, South Korea’s AI policy has pursued two goals: fostering an internationally competitive AI industry and promoting trust in an AI-based society. The Presidential Committee on AI National Strategy, launched on September 8, 2025 with eight divisions, holds formal decision-making and implementation authority—a significant upgrade from the advisory-only committee under the previous administration. In December 2025, it announced a preliminary ROK AI Action Plan comprising 12 strategic areas and 98 specific tasks, with a final version expected in 2026. While comprehensive and ambitious, the Action Plan gives comparatively less attention to democracy than to AI industry development and public-sector AI transformation (AX), yet it still reflects a distinctive social vision for AI.

AI Basic Society

South Korea’s approach to AI governance employs terms such as trust, transparency, accountability, and inclusiveness rather than “democracy” per se, reflecting a particular framing of AI governance values. How countries govern AI to protect or innovate democracy naturally varies.

“AI Basic Society” (ABS) is President Lee’s overarching AI policy vision: an inclusive digital capacity society in which all citizens—regardless of age, occupation, region, or socioeconomic background—are able to understand and utilize AI. The Action Plan describes ABS as a new social compact, committing the government to build an “AI for All” platform that enables citizens to easily understand AI and AI-driven change, directly experience its applications, and naturally integrate AI into their daily lives. A Nationwide AI Capability Transformation Project is also proposed to provide standard models of basic AI competencies and ensure equitable access to AI learning opportunities.[3]

Literally, a “Basic Society” is one where citizens’ basic livelihoods are guaranteed through rights to social services and strengthened social safety nets.[4] In this context, ABS is a concept emphasizing equitable access to AI-assisted social services. The Action Plan allocates 16 tasks across 24 government offices under this segment. A ‘Comprehensive Plan for an AI-Based Basic Society for All,’ to be finalized by the second quarter of 2026, “shall aim to enhance citizens’ quality of life and strengthen the social safety net, while integrally incorporating the principles of AI democracy, technological ethics, and social inclusion” across seven key sectors: labor, welfare, education, finance, culture, safety, and environment.[5] Implementation measures include: institutionalizing the legal status of public-interest data and AI; allocating 10–20% of public-sector AX budgets to public-interest AI; setting minimum private-sector participation levels for social innovation; linking government programs to AI social ventures; building public–private–academic data governance; and establishing ethical review procedures for public-interest data.[6]

The MSIT will launch pilot projects for the AI-based basic society and identify necessary legal and institutional reforms by the fourth quarter of 2027, with nationwide services in AI democracy, inclusive finance, and climate risk management to be initiated by the first quarter of 2030. This process will also promote the securing and open access of public-interest data and the establishment of participatory governance mechanisms.

AI-Assisted Public Sphere

The South Korean government has also proposed creating an AI-assisted public sphere to strengthen representation and help legitimize procedural democracy in the AI era. Kreps and Kriner identify AI threats to democracy in three key areas: representation, accountability, and trust.[7] Altman highlights risks to direct democracy, arguing that AI-driven astroturfing can distort grassroots movements and undermine deliberation in public forums. Unlike traditional participatory mechanisms with high legal and logistical thresholds, AI-assisted forums can be created with minimal oversight, making it harder for citizens to trust the growing number of such spaces.[8] Government-led AI public forums designed with robust trustworthiness safeguards are therefore worthy of serious exploration.

The Action Plan proposes creating AI public forums and experimental citizen labs. It writes: “The pan-governmental AI public forum shall function as a multi-stakeholder collaborative platform that collects data on the risks and opportunities arising from AI transformation and identifies key agenda items for discussion. At the same time, it shall enable experts, stakeholders, and government officials to deliberate together with citizens and develop these discussions into concrete policy measures.”[9]

South Korea has a strong track record of digital participatory mechanisms—including opinion surveys, new policy initiatives, evaluations, citizen ombudsman systems, and petitions. Government-led AI forums could fundamentally transform this model of citizen engagement: they can be faster and more effective through access to public data and multi-stakeholder participation within a shared digital space. At the same time, they must be carefully designed to address potential representation gaps among participating citizens, particularly along partisan and generational lines. Designating non-partisan NGOs as the coordinating “control tower” of these public forums—supported by government infrastructure and public data—could be a viable approach.

International Cooperation for AI Governance

No global AI governance framework yet exists despite extensive discussions on AI’s benefits and harms. In September 2024, the UN High-Level Advisory Body on AI released its final report, Governing AI for Humanity, recommending seven items as an agile and flexible governing approach.[10] So far, little progress has been made. AI governance structures remain fragmented across different forums, lacking the coherence required to manage rapid AI development and deployment.

Three dominant models currently shape the global landscape. The United States advances a market-driven model led by private innovation and corporate self-regulation, reflecting its broader market-oriented regulatory philosophy. The European Union has adopted a comprehensive regulatory approach grounded in human rights and democratic values, exemplified by the world’s first comprehensive AI Act introduced in 2024. China, by contrast, follows a government-led model rooted in techno-nationalism, prioritizing political stability and social control while regulating AI in targeted sectors and use cases.

For many Asian democracies, none of these models offers a perfect fit. Unlike the US or China, they do not host dominant global AI firms or command the capital to independently build full AI ecosystems. At the same time, they are eager to develop domestic AI capabilities and secure access to advanced technologies, yet—similar to the EU—remain highly sensitive to the societal risks associated with AI deployment. This combination of technological aspiration and risk awareness places them in a structurally distinct position. The central challenge is therefore not to choose among the US, EU, or Chinese models, but to develop an alternative framework that reconciles technological access with robust safeguards. In the Indo-Pacific, this requires transforming existing guidelines into institutionalized cooperation, strengthening regional dialogue, and aligning with global governance processes.

1. AI Diplomacy

As a leading semiconductor manufacturer and advanced digital society, South Korea has actively joined global AI cooperation. It hosted the AI Seoul Summit in May 2024, focused on safety, innovation, and inclusivity. The Summit’s Declaration recognized the importance of interoperability between AI governance frameworks and called for creating or expanding AI safety institutes, research programs, and supervisory bodies. It also called for protecting democratic values, the rule of law, human rights, and fundamental freedoms, and for bridging AI and digital divides between and within countries to advance the UN Sustainable Development Goals.[11]

South Korea has also begun incorporating AI collaboration into its bilateral partnerships. During President Lee’s state visit to Singapore on March 2, 2026, both governments shared the ABS vision and expressed intent to pursue an “AI Cooperation Framework.” While AI cooperation has become a common element in international partnerships, South Korea needs to identify more focused areas for collaboration, as priorities will inevitably vary depending on partner countries’ levels of economic development and digital infrastructure.

The India–Japan AI Initiative offers a useful model in this regard. Strengthened through the Japan–India AI Strategic Dialogue launched in January 2026, it focuses on creating a trustworthy AI ecosystem, advancing large language models (LLMs), and supporting startups to counter technology monopolies and promote safe and ethical AI use. At its core is the joint development of international standards for “Trustworthy AI,” based on principles such as personal data protection, algorithmic transparency, and the elimination of bias—distinct from the approaches taken by US Big Tech or China. The initiative also carries geopolitical significance as a potential “third AI bloc” amid intensifying technological competition.[12]

As Japan advances AI ethics and democratic norms—as reflected in the Hiroshima Process, which established international guiding principles for AI-developing organizations—South Korea should enrich the ABS concept with a concrete operational framework that goes beyond broad vision. AI-related foreign assistance programs could serve as an important avenue for operationalizing this framework. South Korea also needs to step up AI cooperation with ASEAN. In 2025, ASEAN expanded its AI Governance and Ethics Guide to include nine recommendations addressing the risks of generative AI, positioning itself as a potential regional norm entrepreneur.[13] However, for this framework to become a viable governance model, it must be embedded in broader regional cooperation involving major AI players such as Korea, Japan, and Taiwan. With its strengths in AI infrastructure and hardware, South Korea is well positioned to help fill this gap.

2. AI Foreign Assistance

The Action Plan envisions a Global Alliance for an AI Basic Society—an open collaborative platform bringing together government, industry, academia, and civil society—to link international standards on AI ethics, trustworthiness, and fairness; build inclusive data governance; expand public AI capabilities; and promote global joint projects. The Alliance “aims to contribute to Korea’s emergence as a global hub for the AI Basic Society model…a strategic diplomatic initiative to present a Korean model of ‘K-democracy and growth in the AI era’ to the international community by exploring an AI development strategy that is democratic, public-interest-oriented, and growth-driven.”[14]

To translate this vision into concrete programs, KOICA concluded an MOU with the National Information Society Agency (NIA) in February 2026 to provide “AI for All” ODA. Building on prior cooperation—including the dispatch of IT volunteer corps, consulting on development cooperation projects, and expert advisory services—the revised agreement significantly expands the scope to include the AI field. The two institutions plan to cooperate in: promoting AI ethics in international development cooperation; providing expert advisory support on AI policies and regulatory frameworks for developing countries; sharing domestic and international networks and information; and promoting institutional publications.

To make these plans effective, South Korea’s global vision must translate into concrete democratic models that can adapt to diverse local contexts. Because democratic institutions in many developing countries do not function effectively, clear guardrails are needed to ensure that AI-focused ODA genuinely supports democratic governance. At the same time, the idea of “K-democracy in the AI era” remains underdefined. Treating AI merely as an extension of South Korea’s digitally active civic culture risks overlooking its more transformative—and potentially disruptive—impact in the Global South.

Conclusion

Shany’s proposed AI Bill of Rights outlines seven key rights: the right of access to AI systems; the right to privacy-related protections from harmful AI uses; the right to be free from algorithmic bias and unfairness; the right to algorithmic transparency and explainability; the right not to be subject to algorithmic manipulation; the right to human decision-making and human-to-human interaction; and the right to accountability for harms caused by the use of AI systems.[15]

South Korea’s “AI for All” vision aligns closely with the right to access AI systems, though issues of algorithmic transparency and accountability remain less explored. The country’s strengths lie in reducing AI access gaps and enhancing public service effectiveness. Building on its relatively successful digital governance, which has already enabled citizen participation in policy processes, South Korea is now rapidly expanding AI-assisted services in citizens’ interactions with government.

In a context of increasing political polarization, AI-assisted public forums could also produce negative effects by intensifying political confrontation. At the same time, AI platforms designed to broaden participation could help depolarize public discourse and contribute to solving social problems. Efforts to strengthen the deliberative capacity of citizens should therefore involve not only the government but also citizens themselves and other private-sector actors.

South Korea’s international cooperation for responsible AI also deserves attention. While its multilateral initiatives are expanding, they could be complemented by more regionally focused cooperation. Collaboration with Japan, Singapore, and Taiwan could be particularly promising. When engaging with the Global South, South Korea’s AI-related ODA should also clarify how AI assistance can support democratic institutions—moving beyond technical cooperation and training exchanges.■

[1]Aviv Ovadya. 2023. "Reimagining Democracy for AI." Journal of Democracy 34 (4): pp. 162-170.

[2]AI Ethics Communication Channel. N.d. "The National Guidelines for AI Ethics." https://ai.kisdi.re.kr/eng/main/contents.do?menuNo=500011.

[3]Presidential Committee on National AI Strategy. 2025. "ROK AI Action Plan." December. p. 127.

[4]Lee, Jae-myung. 2025. "People Sovereignty Government, Basic Income or Basic Society." Le Monde Diplomatique Korean Edition. August 5. https://www.ilemonde.com/news/articleView.html?idxno=21018. (Accessed: February 10, 2026)

[5]Presidential Committee on National AI Strategy. 2025. "ROK AI Action Plan." December. p. 129.

[6]Presidential Committee on National AI Strategy. 2025. "ROK AI Action Plan." December. p. 130.

[7]Kreps, Sarah and Doug Kriner. 2023. "How AI Threatens Democracy." Journal of Democracy 34 (4). pp. 122-131.

[8]Altman, David. 2026. "The AI Democracy Dilemma." Journal of Democracy 37 (1). pp. 32-44.

[9]Presidential Committee on National AI Strategy. 2025. "ROK AI Action Plan." December. p. 132.

[10]Seven recommendations are: the establishment of an IPCC-like independent international scientific panel on AI to provide authoritative, science-based assessments; the launch of a twice-yearly intergovernmental and multi-stakeholder policy dialogue on AI governance to share best practices and promote common understanding on the implementation of AI governance; the creation of an AI standard exchange bringing together representatives form national and international standard-development organizations, tech companies, and representatives from civil society and the international science panel; the creation of an AI development network to align regional and global AI capacity efforts and build AI governance capacity of public officials; the creation of a global AI fund to put a floor under the AI divide; and the creation of a global AI date framework outlining data-related definitions and principles for global governance of AI training date; and the creation of AI office within the Secretariat. United Nations. N.d. "Governing AI for Humanity: Final Report." https://www.un.org/sites/un2.un.org/files/governing_ai_for_humanity_final_report_en.pdf.

[11]Ministry of Foreign Affairs, Republic of Korea. 2024. "Seoul Declaration for Safe, Innovative and Inclusive AI by Participants Attending the Leaders' Session of the AI Seoul Summit." May 21. https://www.mofa.go.kr/eng/brd/m_5674/view.do?seq=321007. The Summit also stated to strengthen international cooperation on Al governance through engagement with other international initiatives at the UN and its bodies, G7, G20, the Organization for Economic Co-operation and Development (OECD), the Council of Europe, and the Global Partnership on AI (GPAI). The Hiroshima AI Process Friends Group, the updated OECD AI principles, and the recently adopted UN General Assembly resolution “Seizing the opportunities of safe, secure and trustworthy artificial intelligence systems for sustainable development,” the Global Digital Compact in advance of the Summit of the Future in September 2024 were all acknowledged.

[12]Kurihara, Toshihiko and Sukirt Kaur. 2026. "Indo–Japanese Collaboration on Artificial Intelligence." Center for Social and Economic Progress, Executive Policy Brief. January. https://csep.org/wp-content/uploads/2026/01/INDO%E2%80%93JAPANESE-COLLABORATION-ON-ARTIFICIAL-INTELLIGENCE.pdf.

[13]ASEAN Secretariat. 2025. "Expanded Guide: AI Governance and Ethics – Generative AI." https://asean.org/wp-content/uploads/2025/01/Expanded-ASEAN-Guide-on-AI-Governance-and-Ethics-Generative-AI.pdf.

[14]Presidential Committee on National AI Strategy. 2025. "ROK AI Action Plan." December. p. 135.

[15]Shany, Yuval. 2025. "White Paper: The Need for and Feasibility of an International AI Bill of Human Rights." Institute for Ethics in AI, University of Oxford. November. https://afp.oxford-aiethics.ox.ac.uk/sitefiles/white-paper-professor-yuval-shany.pdf.

■
Sook Jong Lee is a Distinguished Fellow at the East Asia Institute and a Distinguished Professor at Sungkyunkwan University.

■ Edited by Jaehyun Im, Research Associate

    For inquiries: 02 2277 0746 (ext. 209) | jhim@eai.or.kr