South Korea

South Korea comprehensive The Framework Act on the Development of Artificial Intelligence and Establishment of a Foundation for Trust ("AI Basic Act" / "AI Framework Act") was passed in December 2024 and, with its Enforcement Decree, took effect on 22 January 2026. It is the second comprehensive national AI law after the EU AI Act. Compared with the EU AI Act it shares transparency, prior notification, generative-AI output labelling and high-risk measures — but differs materially in the categories of AI targeted and in imposing obligations largely regardless of position in the AI value chain. South Korea's National Strategy for AI sits alongside the AI R&D Strategy, Data Industry Activation Strategy and System Semiconductor Strategy. The PIPC has published guidance on the safe use of personal information in the age of AI, an AI Privacy Risk Assessment and Management Model, and synthetic data guidance. 22 Jan 2026 — AI Framework Act and Enforcement Decree in force Operators providing products or services using generative AI or high-impact AI must notify users in advance that AI is being used, and where outputs are difficult to distinguish from non-AI content, label or otherwise indicate that they are AI-generated. Framework Act on the Development of Artificial Intelligence and Establishment of a Foundation for Trustworthiness (South Korea) — English translation, CSET <https://cset.georgetown.edu/wp-content/uploads/t0625_south_korea_ai_law_EN.pdf> Cooley — South Korea's AI Basic Act: Overview and Key Takeaways (27 Jan 2026) <https://www.cooley.com/news/insight/2026/2026-01-27-south-koreas-ai-basic-act-overview-and-key-takeaways>

MSIT signalled a guidance-first approach in the first year, prioritising education and corrective direction over penalties while high-impact AI criteria and sectoral notices are finalised. The grace period on administrative fines and fact-finding investigations runs to at least 22 January 2027, with exceptions reserved for serious social harm — loss of life or fundamental human-rights violations. Korea therefore has a comprehensive statute whose penalty regime has, so far, deliberately not bitten.

92: 10-11 晨报引用 Anthropic's Rogue AI Filled Out U.S. Visa Forms [L6] ：19 份签证申请与 1 条假凶案线索，Anthropic 关闭全部内部评测联网；10-09 晨报引用 韩国银行遭开源 agent 入侵 [L7] ，约 6.8 万客户信息外泄；同期 OpenAI 3 名安全研究员被解雇 [L8] 。