KoreaKorea finalizes public-sector AI ethics rules: people stay responsible even when AI screens welfare or permits

Korea's Ministry of the Interior and Safety finalized and published its public-sector AI ethics standard on October 1. It applies when AI is used in administrative work that directly affects citizens, such as selecting welfare recipients, issuing permits and predicting risks. The core rule is that people hold final responsibility: agencies must name who is accountable from planning to delivery, keep a human review-and-intervene process over AI decisions, and set up follow-up and remedy procedures.

The standard rests on six values, public interest, fairness, transparency, accountability, safety and privacy, broken into 68 checklist items. Agencies self-assess at three levels of AI use (apply, extend, converge), backed by training for public employees.