Framework Act on the Development of Artificial Intelligence and the Creation of a Foundation for Trust

Article 1 Purpose

The Act aims to establish basic requirements for sound artificial intelligence development and trustworthy AI, thereby protecting citizens' rights and interests, improving quality of life, and strengthening national competitiveness.

Article 2 Definitions

Defines artificial intelligence as electronic implementation of human intellectual abilities; AI system as an autonomous system producing outputs affecting environments; high-impact AI as systems with significant impact on human life, safety or fundamental rights in areas including energy, water, health, medical devices, nuclear safety, biometric analysis for criminal investigation, employment and loan decisions, transportation, state agency decision-making affecting citizens, and student evaluation in education. Defines generative AI as systems generating text, images, videos and other outputs by imitating input data structure; AI industry as developing, manufacturing or distributing AI products and services; AI business operators as developers or users of AI; users and impacted persons affected by AI; AI society as one creating value through AI in all fields; AI ethics as standards for safe and trustworthy AI respecting human dignity; and training data as data used for AI development.

Article 3 Basic principles and the State's responsibilities

Requires AI development to enhance safety and trustworthiness; entitles impacted persons to clear and meaningful explanations of AI decision criteria to the extent technically and reasonably possible; directs state and local governments to support AI business operators, help citizens adapt to AI changes, and ensure participation and inclusion of AI-vulnerable groups in policy development.

Article 4 Scope of application

The Act applies to conduct outside South Korea affecting the domestic market or users; does not apply to AI developed and used solely for national defense or national security purposes as prescribed by Presidential Decree.

Article 5 Relationship to other statutes

This Act is the primary law for AI and related matters unless other laws expressly provide otherwise. Future legislation concerning AI must align with this Act's purpose.

Article 6 Formulation of AI master plans

The Science and ICT Minister formulates, modifies and implements triennial AI master plans covering basic policy directions, professional training, AI ethics, funding, training data standards using public data, fairness and transparency, social responses to AI, AI-vulnerable group access, and other matters strengthening national competitiveness, subject to Presidential Council deliberation and after consulting relevant agencies and local governments; minor modifications are exempted from Council approval. The Science and ICT Minister may request relevant agencies, local governments and public institutions to provide data for master plan formulation, which they must comply with unless compelling reasons exist; the plan serves as a sectoral implementation plan for intelligent informatization; central agencies and local governments must consider the plan when establishing policies in their jurisdictions.

Article 7 Presidential Council on National Artificial Intelligence Strategy

Establishes a Presidential Council with up to 60 members including the President as chair to deliberate major AI policies and trustworthiness matters. Members include relevant government agency heads, national security and presidential office officials responsible for AI, and up to 2 years of appointed experts who may serve only one additional term. The Council lasts 5 years from the Act's entry into force.

Article 8 Functions of the Council

The Council must decide on inspecting AI master plans, establishing national AI vision and strategies, coordinating AI policies across ministries, monitoring implementation, developing R&D strategies, identifying regulatory barriers, expanding AI data centers, promoting data collection and utilization, and regulating high-impact AI. It may recommend AI practices and ethics to government agencies and business operators, who must formulate improvement plans.

Article 9 Exclusion of, challenge to, and recusal by member

Members must be excluded from deliberation when they or their family members have a direct interest in the agenda item. Parties may request challenging a member's impartiality, and the Council decides by resolution. Members must recuse themselves if either ground applies.

Article 10 Expert standing committees

The Council may establish specialized expert standing committees, special committees for specific AI issues, and advisory committees of experts to review AI matters. It may also operate a council of chief AI officers to establish and implement major government-wide AI policies. Presidential Decree prescribes composition and operation details.

Article 11 AI Policy Center

The Minister of Science and ICT has discretion to establish an AI Policy Center to manage policy development and international norm-setting. The Center provides technical support for master plans and policy measures, analyzes AI's societal impact, conducts research on trends and legal frameworks, and performs other assigned duties.

Article 12 AI Safety Institute

The Minister of Science and ICT may establish an AI Safety Institute to professionally protect citizens from AI-related risks and build public trust. The Institute researches safety risks and policies, develops evaluation criteria, advances safety technologies, promotes international cooperation, and supports system safety compliance.

Article 13 Support for development and safe use of AI technology

Government may support AI research, development, testing, commercialization, and technology transfer. It may support safe use through emergency stop functions, privacy protection design, social impact assessments, and human rights-respecting AI development. Government must disclose results publicly while allowing protection periods for developers to receive royalties.

Article 14 Standardization of AI technology

Government may establish, revise, and disseminate AI standards and conduct research on domestic and international standards. It may publicly notify standards and recommend compliance. Government may support private sector standardization efforts and maintain cooperation with international standards organizations.

Article 15 Establishment of policies measures related to AI training data

The Minister of Science and ICT must implement policy measures to support production, collection, management, distribution, and quality of training data. The Government may fund eligible projects, establish an integrated training data system available to the private sector, and collect usage fees from system users.

Article 16 Policy measures for adoption and utilization of AI technology

The state and local governments shall establish policies promoting AI adoption by enterprises and public institutions, including development support, consulting, data generation assistance, education for small and medium enterprises, funding, and other measures. Public entities must prioritize designated AI products or services when procuring, unless unsuitable; those making such purchases are not liable for damages unless caused by their intentional conduct or gross negligence.

Article 17 Special support for small and medium enterprises, etc

Small and medium enterprises shall receive priority in AI-related support policies. The government shall promote their participation in the AI industry and reflect this in the master plan. The Minister may provide support for safety assessments and trustworthiness measures.

Article 18 Promotion of business start-up

The government may implement programs identifying and fostering AI entrepreneurs, providing education, supporting technology commercialization, valuing AI technology, offering research outcomes, fostering support institutions, and funding other startup initiatives. Local governments may invest in startup support organizations. Central agencies may provide venture capital funds; these funds shall be created from state or local government contributions, private investor contributions, and other incidental income.

Article 19 Promotion of AI convergence

The government shall establish policies facilitating AI convergence with other industries and promoting cross-sectoral AI utilization. The government may prioritize AI convergence research and development in national programs and actively support temporary permissions and regulatory exceptions for demonstration of convergence products and services.

Article 20 System improvement

The Government shall work to improve laws, regulations, and other systems to promote AI industry development and build public trust. The Government may support research on legal frameworks and gather diverse stakeholder views to facilitate such improvements.

Article 21 Securing AI professionals

The Minister shall implement programs developing and supporting AI professionals through education, employment promotion, career path expansion, international training, welfare improvement, and other support measures. The government may secure overseas professionals through investigation of international experts, establishing networks, employment support, overseas expansion of domestic institutes, and attracting international AI organizations and events. The government shall identify international AI trends, promote cooperation, and may support individuals and organizations through technology exchange, market analysis, joint research, capital attraction, marketing activities, and other competitive strengthening measures.

Article 22 2 (Establishment and support of AI research institutes)

Universities, enterprises, and other entities may establish AI research institutes independently or jointly with ministerial approval, requiring at least three promoters and adequate personnel and facilities to conduct technology research, interdisciplinary development, commercialization, professional training, and international cooperation. The Minister may revoke permission if institutes obtained it fraudulently, became unable to meet their purposes, failed to meet requirements, conducted unauthorized business, violated statutes, harmed public interest, or failed to start operations within 6 months or had no business for 1 year, with mandatory revocation for fraud or impossibility and mandatory hearings before revocation. The government may establish and operate its own AI research institutions where necessary for securing innovative technology.

Article 23 Designation of AI clusters

State and local governments may designate AI clusters grouping enterprises and organizations in AI research and development to foster the industry, providing administrative, financial, and technical support as prescribed. The Minister may revoke designation if obtained fraudulently or if achieving the purpose becomes difficult, with mandatory revocation for fraud. Government may establish dedicated institutions to support regional clustering and contribute or subsidize their expenses.

Article 24 Establishment of AI demonstration infrastructure

The State and local governments may build facilities for testing and evaluating AI technologies developed by business operators. They may also make demonstration infrastructure available to operators through designated institutions to support testing and verification activities.

Article 25 Implementation of policy measures related to AI data centers

Government shall implement policy measures encouraging establishment and operation of AI data centers. Government may provide administrative and financial support for AI data center establishment and operation, support their use by small and medium enterprises and research institutes, and support balanced regional development of AI-related infrastructure.

Article 26 Establishment of Korean AI promotion association

Persons engaged in AI research and practice may create or be appointed as a Korean AI promotion association with authorization from the Minister of Science and ICT to advance AI development and industry; the Association is a corporation that promotes AI technology use, surveys the AI sector, operates shared facilities, supports international expansion, and provides education on trustworthy AI. Public authorities may fund the Association's projects and operations within budgets, and bylaws and Presidential Decree shall govern membership qualifications and further governance.

Article 27 AI Ethics Principles

The Government may issue AI ethics principles covering safety and trustworthiness, accessibility for all users, and AI's contribution to human well-being. The Minister must develop action plans, consult stakeholders, and publicize the principles. The Minister may advise administrative agencies and local governments on consistency between their ethics standards and these principles.

Article 28 Establishment of private-sector self-regulatory AI ethics committees

Educational institutions, research institutes, AI business operators, and other AI-related institutions may establish independent committees to monitor ethics compliance, investigate safety and rights violations, supervise AI development activities, provide ethics education, and develop sector-specific guidelines. Committees must include members of both genders and include external persons with expertise in social and ethical matters.

Article 29 Preparation of policy measures to create foundation for trust in AI

Government shall prepare policy measures to minimize AI risks and create trust, including creating safe environments for AI use, forecasting AI impacts, supporting safety and certification technologies, providing education on AI ethics, supporting business operator safety rules, and promoting private sector cooperation on safety and ethical guidelines.

Article 30 Support for verification and certification of AI safety and trustworthiness

The Minister may implement projects to support voluntary verification and certification activities by organizations, including disseminating development guidelines, supporting research, funding equipment and systems, and training professionals. Small and medium enterprises may receive information and financial support; business operators providing high-impact AI should pursue certification; state agencies using high-impact AI must prefer certified products.

Article 31 Obligation to ensure AI transparency

AI business operators must notify users in advance that high-impact or generative AI is used, and indicate when outputs are AI-generated. When outputs such as synthetic media are difficult to distinguish from real ones, operators must notify users in a readily recognizable manner, except where this would hinder display or enjoyment of artistic or creative works.

Article 32 Obligation to ensure AI safety

AI business operators must implement risk identification, assessment, and mitigation throughout the entire lifecycle for AI systems meeting a compute threshold set by Presidential Decree, and establish risk management systems to monitor and respond to safety incidents. Results must be submitted to the Minister, who shall publicly determine implementation methods.

Article 33 Confirmation of high-impact AI

AI business operators must review whether their AI qualifies as high-impact and may request the Minister to confirm this status. The Minister may establish a specialized committee for advice and shall disseminate guidelines on high-impact AI criteria and examples.

Article 34 Responsibilities of business operators regarding high-impact AI

Operators providing high-impact AI must establish risk management, explanation measures regarding AI results and training data to the extent technically feasible, user protection measures, human oversight, and retain verification documents. Measures equivalent to those prescribed by Presidential Decree satisfy this obligation; the Minister may recommend compliance.

Article 35 Impact assessment of high-impact AI

Operators providing high-impact AI products or services should conduct impact assessments on fundamental rights, considering vulnerabilities of specific groups. State agencies and public entities must preferentially consider products that have undergone impact assessments.

Article 36 Designation of domestic representative

Foreign AI business operators meeting user or revenue thresholds must designate a domestic representative with a Korean domicile or place of business to handle submission of safety results, confirmation requests for high-impact AI, and support for safety measures. The operator is deemed responsible for violations by its representative.

Article 37 Expansion of financial resources for promotion of the AI industry

The State must develop a plan to progressively increase funding to support AI policy implementation. The Minister of Science and ICT may encourage public institutions to fund AI industry programs. The State and local governments must establish conditions for private enterprises and other entities to boost their contributions to AI-related initiatives.

Article 38 Compilation of fact-finding surveys, statistics, and indicators

The Minister of Science and ICT, working with the Minister of Data and Statistics, must develop, maintain and release data on domestic and international AI to support policy planning and implementation. The Minister may ask government agencies, local authorities and public bodies for data cooperation; these entities must comply unless compelling reasons prevent it. Presidential Decree shall establish additional requirements for producing, managing and disclosing these surveys and indicators.

Article 39 Delegation of authority and entrustment of tasks

The Minister of Science and ICT or heads of central administrative agencies may delegate authority to subordinate agencies or regional mayors/governors, who may further redelegate to Si, Gun, or Gu heads. The Government may entrust to relevant institutions and organizations the implementation of AI technology development support, training data projects, integrated provision systems, business start-up support, verification and certification, statistics compilation, and other AI industry and ethics tasks, as prescribed by Presidential Decree.

Article 40 Fact-finding investigations

The Minister of Science and ICT may request data submissions or conduct inspections when violations of Articles 31, 32 or 34 are discovered, suspected, or reported by complaints. The Minister may authorize officials to access AI business operator premises to examine records and materials. After investigation, the Minister may direct violating operators to cease or remedy breaches.

Article 41 Legal fiction as public officials for purposes of applying penalty provisions

Council members who are not public officials are deemed public officials for purposes of applying Criminal Act penalty provisions (Articles 127 and 129-132). Executive officers and employees of institutions or organizations entrusted with tasks under this Act are likewise deemed public officials for applying the same Criminal Act provisions.

Article 42 Penalty provisions

A person who discloses secrets learned in the course of duties or uses them for purposes other than duty-related ones, violating Article 7(9), shall be punished by imprisonment with labor not exceeding 3 years or a fine not exceeding 30 million won.

Article 43 Administrative fines

Administrative fines not exceeding 30 million won are imposed on persons who fail to provide required notification, fail to designate a domestic representative, or fail to comply with violation correction orders. The Minister of Science and ICT imposes and collects these fines as prescribed by Presidential Decree.