As the preparation of the subordinate legislation for the 'Basic Act on the Promotion of Artificial Intelligence (AI) and the Formation of Trust' (AI Basic Act), which is set to take effect in January next year, has been delayed by more than a month, voices are increasing that the enforcement of its regulatory provisions should be postponed by several years. There are concerns that hastily implementing the AI Basic Act may hinder industrial development, as domestic laws and systems are unable to keep pace with the rapid advancement of technology.
Some experts and industry insiders argue that a grace period of 2 to 3 years should be given to prepare the subordinate legislation of the AI Basic Act, allowing the domestic AI industry time to grow without unnecessary regulations. In particular, as the competition in AI among major countries like the United States and China becomes fierce, it is crucial not to miss the anticipated 'golden time' of the next 2 to 3 years for the government to achieve its goal of becoming one of the 'three major AI powers.'
According to the Ministry of Science and ICT and the AI industry on the 5th, the government postponed a meeting scheduled for this month on gathering opinions regarding the enforcement decree, notice, and guidelines that contain the core content of the AI Basic Act set for implementation. The delay was attributed to the fact that internal coordination regarding the subordinate legislation has not yet been completed. Initially, the Ministry aimed to finalize the subordinate legislation for the AI Basic Act by last month and then announce a draft of the enforcement decree, but this has been postponed for more than a month.
The AI Basic Act, following the European Union's AI Act, is the second AI-related legislation enacted worldwide, passing the National Assembly in December last year and set to be fully enforced on January 22 next year. It is largely composed of fostering the AI industry and establishing a trust-based foundation (risk management), and the AI industry has argued that the regulatory standards have been unclear and require improvement.
A major issue is the definition of 'high-impact AI.' The AI Basic Act distinguishes between 'high-impact AI,' which significantly affects life, safety, and fundamental rights, and 'general AI,' imposing prior notification and examination and certification obligations on high-impact AI operators. The industry has pointed out that the criteria for high-impact AI are excessively vague and broad.
The regulation requiring labeling of AI-generated content is also identified as a provision that needs modification by the industry. Operators are required to indicate the source (watermark) on results such as deepfakes created using generative AI and must notify this, yet there are concerns in the industry that such measures could hinder content quality and creative activities.
With less than six months until the implementation of the AI Basic Act, the industry believes it will be difficult to prepare effective enforcement decrees that reflect detailed requests from the sector, stating that subordinate legislation should be supplemented over time. In particular, as the government has selected five teams for the project to develop a 'sovereign AI foundation model' representing Korea and has begun the development of sovereign AI, they argue that regulations hindering the creation of the domestic AI ecosystem should be minimized.
Park Sang-cheol, a professor at Seoul National University Law School, noted, 'The current bill mostly classifies most generative AIs as high-impact AI, imposing regulation that could severely obstruct industrial development,' emphasizing that the grace period for the AI Basic Act should be actively considered. He suggested that a regulatory system considering the characteristics of each industry should be established over time. Professor Park explained, 'Recruitment AI and autonomous driving AI require different considerations and should be categorized accordingly.'
Countries leading in AI, such as the U.S., are unveiling policies focused on promotion. U.S. President Donald Trump recently announced the 'AI Action Plan,' accelerating the relaxation of regulations. Trump stated, 'AI is a competition that the U.S. must win,' vowing to devote full efforts to fostering the industry. Recently, there has been growing sentiment in the EU that instead of hastily enforcing AI laws, they should be carefully supplemented.
There are also opinions suggesting that instead of introducing a comprehensive AI Basic Act, Korea should adopt a 'tailored regulation' approach that integrates AI into the existing legal framework, similar to major countries such as the U.S., the U.K., and China.
The government plans to establish a flexible legal framework focusing on industrial promotion. An official from the Ministry of Science and ICT said, 'We plan to gather industry opinions this month and announce the draft enforcement decree around September.' Regarding the industry's request for a grace period for the AI Basic Act, he stated, 'We are considering a plan to operate a regulatory grace period.'