South Korea - Personal Data Protection (19234/2023)
Personal Information Protection Act (amendment including provisions on automated decision-making/AI)
개인정보 보호법 (자동화된 의사결정/AI 관련 조항 포함 개정)
South Korea
RAI-KR-NA-PIPAIXX-2023South Korea - Personal Data Protection (19234/2023) is In Force in South Korea as of 9 Sep 2026, according to elaw.klri.re.kr.
ActData Protection and PrivacyTransparency and DisclosureGovernance and OversightEstablishes data-subject rights to explanation or refusal of automated AI decisions and mandates transparency duties for data controllers under Act No. 19234, adopted by South Korea's National Assembly in 2023. Enforced by the Personal Information Protection Commission, the amended statute came into force on 2024-03-15 and remains In Force.
Summary
Background and Purpose: The Personal Information Protection Act (PIPA) of the Republic of South Korea underwent a major amendment adopted by the National Assembly on 14 March 2023 (Act No. 19234). The amendment responds to the rapid growth of data-driven services and artificial intelligence (AI) and aims to strike a balance between enabling data use for innovation and strengthening protection of individual rights. Many of the delegated details were implemented through the Enforcement Decree (Presidential Decree No. 34309) and administrative guidance issued by the Personal Information Protection Commission (PIPC).
Automated decision-making and AI-specific provisions: One of the most significant novelties is the formal recognition of data-subject rights in relation to decisions produced by fully automated systems (including AI). The amendment introduces a right for data subjects to request a concise, meaningful explanation or human review of decisions made through a fully automated process that materially affects their rights or obligations; data subjects may also refuse such automated decisions in cases with significant impact. Controllers are required to disclose, in advance (for example through privacy policies), the fact that automated decision-making is used and the criteria and processing procedures that will be applied, in a form that is accessible and intelligible to data subjects.
Data portability and data-subject empowerment: The amendment expands the right to data portability, allowing data subjects to request transmission of their personal information to themselves or to third-party controllers. The Enforcement Decree and subsequent PIPC guidance specify the procedures, formats, permissible limits, and exceptional grounds for refusal or suspension.
Governance, accountability and organisational obligations: The amendment strengthens organisational responsibilities, requiring larger controllers and certain public institutions to appoint Chief Privacy Officers (CPOs) who meet prescribed experience and qualification thresholds (e.g., minimum years of experience). Controllers must adopt internal management plans, implement technical and organisational measures (including de-identification, pseudonymization, access control, and logging), perform privacy impact assessments where appropriate, and keep records of processing.
Cross-border transfers and transparency: The law increases transparency obligations for overseas transfers: privacy notices must indicate the legal basis for transfers, the categories of recipients, and the countries where processing occurs. Administrative guidance and PIPC materials provide practical steps and examples for safe use of publicly available data in AI training, acceptable measures for de-identification, and risk assessment models.
Enforcement and sanctions: The amended regime enhances administrative enforcement powers, emphasizes corrective orders and fines over certain criminal punishments, and grants regulators (notably the PIPC) expanded oversight authority, including privacy-level assessment powers for public institutions and large controllers. The Enforcement Decree sets administrative procedures, thresholds, and timelines for specific provisions.
Regulatory guidance and AI governance: Following the amendment and the Enforcement Decree, the PIPC has published AI-focused materials — including a July 2024 guideline on processing publicly available personal information for AI development and later guidance addressing generative AI and other AI-specific risks — to clarify legal bases (including legitimate interest), de-identification standards, and lifecycle governance practices.
Implications for actors: The amendment creates concrete obligations for controllers, processors, and foreign entities targeting Korean data subjects. It raises compliance costs for AI projects that process personal data at scale, but also provides clearer rules to enable lawful innovation when controllers can demonstrate governance, risk mitigation, and respect for data-subject rights. Data subjects gain strengthened remedies and new practical rights against opaque automated decision-making.
Full article
Read full text ↗Overview
The March 14, 2023 amendment to the Personal Information Protection Act (PIPA) modernises South Korea's core privacy law to address large-scale data processing and AI-driven automated decision-making. The amendment (Act No. 19234) and its implementing measures (notably the Enforcement Decree and PIPC guidance) create new data-subject rights—including the right to refuse fully automated decisions, and to request explanations or human review—and expand data portability and transparency requirements. The law also ramps up organisational responsibilities (CPO qualification, internal management plans, mandatory security measures) while delegating technical detail and operational rules to the Enforcement Decree and regulatory guidance. Official texts are available from the Korean Law Information Center and the government's privacy portal (see PERSONAL INFORMATION PROTECTION ACT (KLRI)) and implementing decrees at the national law portal (Enforcement Decree (law.go.kr)).
Definitions
PIPA defines key terms relevant to automated decision-making and AI: "personal information" (identifiable natural person data), "sensitive information" (민감정보), "processing" (collection, use, provision, etc.), and for the first time the framework recognises "automated decision" or "fully automated process" meaning a decision that affects a data subject's rights or obligations made by a system operating without substantive human intervention. The Enforcement Decree and PIPC guidance provide interpretive detail about what constitutes 'significant impact,' the scope of automated decision-making, and the boundary between automated and human-in-the-loop systems.
Governance and Institutional Framework
The PIPC remains the principal regulator responsible for supervision, guidance, and enforcement; it has been given expanded powers to assess privacy-management levels of public institutions and large private controllers, and to publish sectoral guidance. Controllers meeting size or processing thresholds must designate a Chief Privacy Officer (CPO) who satisfies qualification criteria; transitional provisions allow a grace period for certain entities to meet the enhanced CPO qualification rules. The government’s privacy portal hosts PIPC guidance documents including the July 2024 "Guideline for Processing Publicly Available Personal Information for AI Development" (see PIPC AI/Public Data Guidance (privacy.go.kr)), and the Enforcement Decree (Presidential Decree No. 34309) sets procedural and threshold rules for implementation.
Key Focus Areas
Automated decision-making: data subjects may request explanations or human review and may refuse fully automated decisions that significantly affect rights/obligations. Controllers must inform subjects when automated decisions will occur and publish decision criteria and processing procedures in accessible formats. Data portability: subjects can request transmission of their data to themselves or third parties; the Decree defines the methods, formats, permissible exceptions and refusal grounds. De-identification & pseudonymization: the amendment strengthens provisions enabling lawful use of de-identified or pseudonymized data for research and AI training while emphasizing technical safeguards and risk assessment. Cross-border transfers & transparency: controllers must disclose the legal basis for overseas transfers and the destination countries. Accountability & security: enhanced internal management plans, privacy impact assessments, access controls, and logging are required for entities processing significant volumes of personal or sensitive information.
Implementation Framework
Implementation is staged and partly delegated: many amendment provisions were effective on 15 September 2023, while the provisions on automated decision-making, data portability and certain technical/delegated matters became effective on 15 March 2024 with enactment of the Enforcement Decree (see KLRI e-book record). The Enforcement Decree establishes numeric thresholds (e.g., daily averages of subject counts), methods for notification, procedures for refusal/review requests related to automated decisions, and transitional measures for CPO qualification. The PIPC has issued practical guidance and templates to support compliance and to interpret delegated rules.
Monitoring and Evaluation
The PIPC is authorised to conduct privacy-level assessments for public institutions and large controllers, to require corrective measures, and to publish guidance and assessment outcomes. The law and Decree mandate recordkeeping and periodic internal reviews; PIPC guidance recommends privacy impact assessments (PIAs) and risk-model frameworks for AI projects and suggests periodic re-evaluation of deployed models to monitor accuracy, bias, and security risks such as model inversion or data leakage.
Penalties, Liability, and Appeals
The amended PIPA strengthens administrative enforcement and emphasizes corrective powers, administrative fines, and remedial orders for violations; it also clarifies liability for misuse of personal data in automated decision contexts. The regime focuses on proportionate administrative penalties and remediation, though criminal sanctions may remain for specific offences. Affected entities retain rights of appeal against administrative orders to administrative courts; the PIPC’s procedural rules and the Enforcement Decree set detailed timelines and remedy mechanisms.
Relationship to Other Instruments
PIPA interacts with sectoral laws (e.g., Act on Promotion of Information and Communications Network Utilization, Medical Services Act, Financial sector AI guidance) and international rules. The PIPC’s AI-specific guidance and industry-specific regulators’ rules (e.g., Financial Services Commission guidance) operationalise sectoral expectations. The amendment anticipates future related legislation such as South Korea’s AI Basic Act and complements standards for data protection and model safety.
International Alignment
South Korea’s amendments borrow concepts similar to other modern privacy regimes—data portability and protections against fully automated decisions comparable to the EU's GDPR—while tailoring rules to local administrative structures and innovation policy. PIPC guidance explicitly references international good practices and seeks interoperability for cross-border data flows, while establishing South Korea-specific thresholds and administrative procedures to govern foreign entities targeting Korean data subjects (see PIPC materials on foreign controllers).
Implementation Timeline
| Event | Date |
|---|---|
| National Assembly adoption (Act No. 19234) | 2023-03-14 |
| Partial provisions effective (first tranche) | 2023-09-15 |
| Enforcement Decree published (Presidential Decree No. 34309) and automated decision/data portability effective | 2024-03-15 |
| Additional delegated measures, guidance and PIPC guidelines (AI/public data guidance) | 2024-07-17 (guideline published) |
Sources and References
| Source | Type |
|---|---|
| PERSONAL INFORMATION PROTECTION ACT (KLRI) — Act No. 19234 (2023) | Primary Source |
| Enforcement Decree (Presidential Decree No. 34309) — law.go.kr | Primary Source |
| PIPC: Guideline for Processing Publicly Available Personal Information for AI Development (2024.7) | Primary Source |
Requirements for a company
What an organisation has to do under South Korea - Personal Data Protection (19234/2023), at a glance. Not legal advice — the table below gives the provision and deadline for each item.
Must do
8- Inform data subjects before making automated decisions and publish decision criteria and processing procedures in accessible formats.Personal information controllers using automated decision-making systems
- Allow data subjects to refuse fully automated decisions that significantly affect their rights or obligations.Personal information controllers using fully automated decision systems
- Provide explanations and human review upon request for automated decisions affecting data subjects.Personal information controllers using automated decision-making systems
- Transmit personal data to subjects or designated third parties upon request using compliant methods and formats.Personal information controllers handling data portability requests
- Designate a Chief Privacy Officer who satisfies official statutory qualification criteria.Personal information controllers meeting size or processing thresholds
- Disclose the legal basis and destination countries for all overseas transfers of personal information.Personal information controllers transferring data internationally
- +2 more in the table below
Must not do
0Nothing in this category.
Should do
1- Perform privacy impact assessments and periodically re-evaluate deployed AI models to monitor bias and security risks.Organisations developing or deploying AI models using personal information
Should not do
0Nothing in this category.
Who must do what
The obligations under South Korea - Personal Data Protection (19234/2023), most serious first. Not legal advice — verify against the official text before relying on it.
| # | Who | Requirement | By when | Where | Severity |
|---|---|---|---|---|---|
| 1 | Personal information controllers using automated decision-making systems | Inform data subjects before making automated decisions and publish decision criteria and processing procedures in accessible formats. “Controllers must inform subjects when automated decisions will occur and publish decision criteria and processing procedures in accessible formats.” | Mar 15, 2024 | — | Critical |
| 2 | Personal information controllers using fully automated decision systems | Allow data subjects to refuse fully automated decisions that significantly affect their rights or obligations. “data subjects may request explanations or human review and may refuse fully automated decisions that significantly affect rights/obligations.” | Mar 15, 2024 | — | Critical |
| 3 | Personal information controllers using automated decision-making systems | Provide explanations and human review upon request for automated decisions affecting data subjects. “data subjects may request explanations or human review” | Mar 15, 2024 | — | Critical |
| 4 | Personal information controllers handling data portability requests | Transmit personal data to subjects or designated third parties upon request using compliant methods and formats. “subjects can request transmission of their data to themselves or third parties” | Mar 15, 2024 | — | Critical |
| 5 | Personal information controllers meeting size or processing thresholds | Designate a Chief Privacy Officer who satisfies official statutory qualification criteria. “Controllers meeting size or processing thresholds must designate a Chief Privacy Officer (CPO) who satisfies qualification criteria” | — | — | Critical |
| 6 | Personal information controllers transferring data internationally | Disclose the legal basis and destination countries for all overseas transfers of personal information. “controllers must disclose the legal basis for overseas transfers and the destination countries.” | — | — | Critical |
| 7 | Controllers processing significant volumes of personal or sensitive data | Maintain internal management plans, access controls, and logging for processing personal or sensitive data. “enhanced internal management plans, privacy impact assessments, access controls, and logging are required” | — | — | Critical |
| 8 | Personal information controllers training AI models with pseudonymized data | Implement technical safeguards and conduct risk assessments when using pseudonymized data for AI training. “lawful use of de-identified or pseudonymized data for research and AI training while emphasizing technical safeguards and risk assessment.” | — | — | Important |
| 9 | Organisations developing or deploying AI models using personal information | Perform privacy impact assessments and periodically re-evaluate deployed AI models to monitor bias and security risks. “suggests periodic re-evaluation of deployed models to monitor accuracy, bias, and security risks” | — | — | Recommended |
Related Regulations
South Korea - Automated Decision Rights Guide
South Korea92% similar
South Korea - AI Impact Assessment
South Korea92% similar
South Korea - AI Privacy Risk Management
South Korea91% similar
South Korea AI Trust Framework Act
South Korea91% similar
South Korea - Public Data Processing Guide
South Korea89% similar
More AI regulation in South Korea
AI regulation in South Korea: full overview
- South Korea AI Security Guide
- South Korea AI Action Plan 2026-2028
- South Korea AI Regulation Overview
- South Korea - Seoul - AI Administration Plan (SAAPXXX/2024)
- Seoul AI Regulation Summary
- South Korea - AI Industry Promotion (AIPTRXX/2022)
- South Korea - AI Responsibility Act
- South Korea - AI Development Framework
© Regulations.AI · updated on 20 Sep 2026 · reviewed against official sources on 9 Sep 2026 using Gemini 3.6 Flash