South Korea - AI Development and Trust (2021)

Overview

The AI Promotion and Trust-Building Bill (original Korean title: 인공지능 육성 및 신뢰 기반 조성 등에 관한 법률안) was a member-drafted bill introduced by Representative Jeong Pil‑mo on 2021-07-01 (의안번호 제2111261호). The proposal sought to establish a national framework to promote the development of the artificial intelligence industry in the Republic of Korea while creating legal duties and transparency measures intended to build public trust. Its stated aims combined industrial promotion (standards, human resources, research support, and coordination) with trust-building measures (ethical principles, transparency, notice, and reporting duties for higher‑impact AI systems). The draft framed AI policy as requiring both promotion and trust-building and proposed national planning, an oversight body, ethics guidance, and a risk-based approach to special-purpose or high-impact AI. The draft was published on lawmaking portals and discussed in committee stages; a stakeholder draft copy is referenced on an industry page for public review (see the full-text/stakeholder submission link below). Official committee and legislative records discuss the proposal and its role in later consolidation into broader AI basic‑law activity. (See details at https://opinion.lawmaking.go.kr and the national statute repository at https://www.law.go.kr.)

Definitions

The 2021 draft contained statutory definitions intended to delineate the scope of covered systems and to trigger heightened duties for certain categories. Key definitional elements in the draft included: definitions of 'artificial intelligence' and 'AI system' focusing on systems that perform learning, inference, perception, judgment, or language understanding; and a category referred to as “special-purpose AI” (special-use or high-impact AI) intended to cover AI whose outputs materially affect life, bodily safety, dignity, or other core interests. The draft delegated precise technical thresholds and implementation details (including technical criteria for classification) to subordinate implementing rules, ministerial guidance, and technical standards rather than fixing detailed technical thresholds in the primary text. For exact draft wording and definitional detail, the original proposal text and the lawmaking portal entries are authoritative sources. (See https://opinion.lawmaking.go.kr.)

Governance and Institutional Framework

The bill proposed a multi-layer governance framework centered on national-level planning and coordination. It envisioned a national AI council or similar central advisory/review body (described in the draft as an 'AI society committee' or equivalent) to review and deliberate basic plans and implementation measures. The Ministry of Science and ICT (or an equivalent central authority identified in the bill) was charged with formulating 3‑year basic plans, coordinating standardization, promoting research and industry support, overseeing ministerial actions, and leading certification and recognition activities for private-sector ethics committees and research centers. The draft also authorized support for research centers, a public-private coordination mechanism, and administrative structures for oversight and standard-setting. The governance design emphasized coordination across ministries, use of ministerial guidance and subordinate rules, and administrative certification/recognition schemes rather than judicial or criminalized enforcement as the primary compliance pathway.

Key Focus Areas

• Statutory purpose and policy balance: Establish a legal framework to promote AI industry growth while ensuring ethical and human-centered development.
• Ethics and human-centered principles: Require AI to respect human dignity, privacy, and ethical norms and authorize publication of ethics guidance.
• Transparency and explainability: Impose duties of notice and explanation for certain special-purpose AI and grant affected individuals rights to information about AI's role in decisions.
• Prior-notification and reporting: Require prior-notification/registration or reporting to the competent ministry for development or provision of certain high-impact AI systems.
• Risk-based approach and standards: Delegate technical thresholds, risk assessment criteria, and certification schemes to ministerial guidance and technical standards.
• Support for industry and standardization: Empower the Ministry of Science and ICT to lead standardization, promotion activities, research support, and human-capacity building.
• Administrative compliance mechanisms: Emphasize administrative oversight, registration/certification, public-private coordination, and corrective measures rather than detailed criminal penalties.

Implementation Framework

The draft proposed an implementation approach combining statutory framework provisions with subordinate rulemaking. Key elements included: ministerial authority to publish ethics guidance and technical standards; three‑year national basic plans formulated and coordinated by the Ministry of Science and ICT; certification or recognition schemes for private autonomous ethics committees and entities operating under approved governance structures; administrative pre-notification or registration systems for special‑purpose AI deployments; recordkeeping and documentation requirements to enable explanation and accountability; and support measures for research centers and industry promotion. The bill envisioned the executive to translate statutory principles into concrete technical standards, compliance checklists, and administrative processes through implementing rules and ministerial directives. Where the primary text set principles and duties, the draft expressly delegated detailed thresholds, certification criteria, and sanctioning mechanisms to subordinate rulemaking to preserve flexibility for technical evolution.

Monitoring and Evaluation

Monitoring and supervisory mechanisms in the draft emphasized administrative oversight and evaluation by competent authorities. The bill provided the government with powers to investigate compliance with notice and reporting duties, to require corrective measures, and to evaluate and certify private-sector ethics committees and recognized bodies. The draft envisaged periodic review through the national basic plan cycle (3‑year plans) and deliberation by the national AI council or equivalent body to assess implementation, update guidance, and coordinate inter-ministerial actions. Detailed monitoring metrics, evaluation methodologies, and reporting formats were left to ministry guidance and implementing rules.

Penalties, Liability, and Appeals

The 2021 draft prioritized supervisory and administrative remedies over a detailed criminal penal code. It focused on administrative enforcement powers—investigation, corrective orders, registration revocation, certification withdrawal, and administrative sanctions—rather than establishing an extensive set of criminal penalties in the primary text. The bill provided for powers to investigate compliance with notice/report duties and to require remedial action; it left specific penalties, fine schedules, and appellate procedures to be specified in implementing rules or subsequent legislative amendments. As drafted, the framework emphasized administrative oversight, certification, registration, and corrective administrative measures as primary enforcement pathways.

Relationship to Other Instruments

The proposed law was drafted as a sectoral framework intended to operate alongside and interact with existing privacy, administrative, and sectoral safety laws in Korea. The bill explicitly anticipated cross-cutting interaction with the Personal Information Protection Act and with sectoral safety and administrative regulations. It provided authority for standards and certification regimes to be implemented through subordinate rules, and it was positioned to complement rather than supplant existing legal instruments. Over time, the draft and other parliamentary AI bills were consolidated into broader AI basic‑law legislation; many objectives of the 2021 draft were carried forward, absorbed, or superseded by subsequent consolidated enactments and subordinate rulemaking.

International Alignment

The 2021 draft itself was primarily a domestic sectoral framework and did not set out comprehensive provisions expressly aligning Korea’s regulatory approach with any single international instrument. The draft’s emphasis on ethics guidance, standardization, risk‑based assessment, and certification is consistent with common international approaches promoted by international organizations and standards bodies; however, the primary text left technical alignment, specific standards adoption (e.g., ISO or other international standards), and cross-border regulatory coordination to subordinate rules and ministerial initiatives. The draft therefore provided a domestic legal foundation from which alignment with international standards and cooperation could be pursued through implementing rules and policy coordination.

Implementation Timeline

Date	Event
2021-07-01	Bill proposed (Representative Jeong Pil‑mo) — 의안번호 제2111261호 was filed on 1 July 2021.
2021-07-02	Bill referred to the Science, ICT, Broadcasting and Communications Committee (referred on 2 July 2021).
2023-02-14	Bill reported / passed in law‑making subcommittee stages and reported in committee-level proceedings (reported press coverage and committee activity in 2022–2023).
2024-11-26	Related consolidated AI legislative activity (AI Basic Act family) culminated in later Assembly activity and public rulemaking discussions; multiple AI bills including this draft were consolidated into broader AI basic‑law initiatives (consolidation and later enactment activity occurred in late 2024–2025).
2025-12-15	Status: Ended — original proposal lapsed or was absorbed into later legislative consolidation and committee alternatives; the standalone 2021 draft did not reach final enactment as introduced.

Compliance Checklist

Requirement	Description
Notice to affected persons	Providers of special-purpose (high-impact) AI must give advance notice to persons affected by AI decisions, as specified for the relevant category of system.
Explanation and access rights	Affected individuals are granted rights to explanation and access to information describing the AI’s role; providers must explain decision-making principles and outcomes on request.
Prior-notification / registration	Certain special-use AI systems require prior-notification or registration with the competent ministry before development or provision; details and thresholds are set in implementing guidance.
Adherence to ethics guidance	Operators must adhere to published ethics guidance and are expected to implement technical and managerial measures for safe development and operation.
Documentation and recordkeeping	Maintain documentation enabling explanation and accountability (data provenance, model descriptions, decision logs) to support audits and explanations.
Certification and recognition	Entities operating private autonomous ethics committees or participating in certification schemes may seek ministry evaluation/certification under schemes authorized by the bill.

Sources and References

Source	URL
National lawmaking portal — bill details and committee records	https://opinion.lawmaking.go.kr/gcom/nsmLmSts/out/2111261/detailRP
National statute repository / official law list (reference to later consolidated AI basic-law enactment)	https://www.law.go.kr/LSW/lsRvsRsnListP.do?chrClsCd=010202&lsId=014820&lsRvsGubun=all&utm_source=openai

Full text / stakeholder submission with draft (industry-hosted reference): https://www.bsa.org/kr/policy-filings/ingongjineung-yugseong-mich-sinloe-giban-joseong-deunge-gwanhan-beoblyulan11261e-daehan (industry-hosted draft copy referenced in the original materials; excluded from the official Sources table above in accordance with official-source listing rules).