What happened
Around 1 October 2026, South Korea's Ministry of the Interior and Safety finalised the 'Public Sector AI Ethics Standards' after drafting in November 2025 and expert review. The standards set six core values (public interest, fairness, transparency, accountability, safety, privacy protection) plus 68 self-check items mapped to three levels of AI usage (basic application, decision support, autonomous decision). The accountability principle codifies that final responsibility/control over AI-assisted administrative decisions rests with humans, requires responsible parties and human review/intervention, and mandates post-management and remedy mechanisms where AI affects rights (e.g. welfare eligibility, licensing, risk prediction).
Why it matters
Authoritative Korean government guidance binding public agencies' AI use, operationalising human-oversight expectations for AI in government decision-making touching citizens' rights — a benchmark for vendors selling into the Korean public sector and the broader Korean AI-governance direction alongside the amended PIPA now in force.
Action needed
Vendors deploying AI for Korean public agencies should align with the six values and 68 self-check items; public agencies will use the tiered self-assessment framework when procuring or operating AI.