South Korea will be the first country in the world to implement an AI law

South Korea prepares to become the first country in the world where comprehensive artificial intelligence legislation will actually come into effect.

As reported by BAKU.WS with reference to The Korea Times, the national framework law "On AI Development and Building a Trust Foundation" will come into force on January 22.

Meanwhile, the European Union, which developed its own AI Act earlier than Seoul, decided to slow down its practical implementation.

The AI law in South Korea was adopted by the National Assembly at the end of last year and officially published in January of this year. Formally, it comes into force exactly one year after promulgation - January 22, 2026.

Until that moment, the government will complete the public discussion procedure for bylaws, which should determine the specific mechanisms for applying the new norms. It is these bylaws that cause the greatest concern for businesses: their final approval is expected practically on the eve of the law's launch, which, according to companies, leaves too little time for adaptation.

The Korean basic law on AI is a framework and covers almost the entire spectrum of modern artificial intelligence technologies. In it, AI is defined as an electronic implementation of human intellectual abilities - from learning and reasoning to perception and language understanding. The concepts of AI systems, generative AI, and high-risk technologies are separately introduced, which automatically includes most existing and developing services in the scope of the law.

However, the document cannot be called exclusively restrictive. Unlike the European approach, which emphasizes risks and sanctions first and foremost, the Korean law simultaneously establishes the state's obligation to support industry development. It provides for the creation of a national AI committee, the development of three-year basic plans, measures to adapt citizens to the spread of technology, and the formation of ethical principles for their application.

A separate feature of the Korean approach was the provision to review legislation every three years, which should allow for prompt adaptation to technological and international changes.

However, it is the practical requirements for businesses that have become the main source of tension. The law introduces mandatory certification of reliability and security, requirements for transparency of datasets used, a registration and verification regime for high-risk AI systems, and mandatory marking of content created with AI assistance.

The last point - about watermarks and other designations for images, videos, and audio - is causing a particularly heated debate. Industry representatives point out that even AI content is often created with the participation of large teams of people, and mandatory labeling "generated by AI" can negatively affect its perception by consumers. Additional questions are raised by the lack of clear criteria for exactly how such marking should be carried out.

Business concerns are also confirmed by surveys. According to a Startup Alliance study, 98% of Korean AI startups do not yet have a full-fledged system for compliance with the requirements of the new law. Nearly half of those surveyed admitted they poorly understand specific regulatory expectations, and almost as many stated that the preparatory measures taken are clearly insufficient. Warnings are being voiced in the industry that if current deadlines are maintained, some services may be forced to urgently change their operating format or even temporarily suspend activities after January 22.

In any case, South Korea's approach strongly contrasts with the European Union. The EU was the first to develop a comprehensive "AI Act", but in the fall announced the postponement of the application of key standards, primarily concerning high-risk AI systems, to a later date - up to 2027.

At the same time, Brussels has moved to ease requirements in the field of personal data protection. The reasons for this step are officially explained by fears that overly strict regulation could weaken the position of European companies in competition with American and Chinese technology giants.

In Seoul, on the contrary, they decided not to postpone the launch of the legal mechanism, despite criticism from the industry. At the same time, the government is trying to smooth out the tension. The Ministry of Science and Information and Communication Technologies of the Republic of Korea has already stated that for at least the first year after the law comes into force, the emphasis will be not on fines and sanctions, but on explanatory work and "soft implementation" of the norms. Administrative penalties will effectively be postponed, and 2026 is seen as a period of adaptation and consolidation of the new system.

Thus, South Korea is the first in the world to move from declarations to practical legal regulation of artificial intelligence. This step makes the country a kind of global testing