South Korea rolls out new law to regulate AI

South Korea has begun enforcing what it describes as the world’s first fully comprehensive legal framework governing artificial intelligence (AI), placing the country at the forefront of global efforts to regulate a technology that is rapidly reshaping everyday life.

The new legislation, known as the AI Basic Act, officially took effect on January 22, 2026, ahead of similar initiatives elsewhere, including the European Union’s AI Act, which will be phased in through 2027.

The South Korean government says the law is designed to promote the adoption of AI while embedding safeguards to ensure public trust and safety. Authorities are planning to launch a guidance platform and a dedicated support centre to help companies navigate the new rules during the transition period.

Oversight and transparency

The new law applies to three areas: high-impact AI, safety obligations for high-performance AI, and transparency requirements for generative AI.

Under the legislation, companies deploying so-called “high-impact” AI systems must ensure human oversight. These systems include fully automated technologies used in sensitive areas such as nuclear safety, drinking water production, transport, healthcare and financial services, including credit scoring and loan screening.

The law also introduces transparency requirements. Companies must inform users in advance when products or services rely on high-impact or generative AI, and clearly label AI-generated content when it could be mistaken for real material.

A key difference from the European approach is how advanced AI models are regulated. While the EU focuses on the risks linked to how AI is used, South Korea relies on technical benchmarks, such as how much computing power is used to train a model. As a result, only a limited number of the most advanced AI systems are expected to face additional safety requirements, according to The Korea Herald.

Penalties and representatives

Failure to comply can result in administrative fines under the new law. A company that does not properly label generative AI content, for example, could face penalties of up to 30 million won (around USD 20,400). Companies will be given a grace period of at least one year before fines are enforced.

By comparison, penalties under the EU’s AI Act are potentially far larger, ranging from 1 percent of global turnover for smaller violations to as much as 7 percent for violations involving banned high-risk uses, according to Reuters.

The law also includes specific provisions for large foreign technology companies operating in South Korea. Global AI service providers with annual worldwide revenue of at least 1 trillion won (around USD 681 million), domestic sales of 10 billion won or more, or at least one million daily users in the country are required to designate a local representative. OpenAI and Google are among the companies that meet those thresholds, according to a recent report by The Korea Times.

The move comes amid a growing international debate over how tightly AI should be governed. The United States has largely favoured a lighter regulatory touch to avoid constraining innovation, while China has introduced targeted rules and floated the idea of a global coordination body for AI governance, as per Reuters.