Draft Reported to AI Strategy Committee, Enactment to Be Completed Within the Year
"Minimum Necessary Regulation" and Guidance Period Before Fines Introduced
In line with the launch of the National Artificial Intelligence (AI) Strategy Committee under the Lee Jaemyung administration, the subordinate regulations of the "AI Basic Act" have established their direction.
Science and ICT Minister Paik Kyung-hoon is holding the 'AI for Science and Technology Meeting' on the 4th at the Korea Energy Research Institute in Yuseong-gu, Daejeon. Provided by the Ministry of Science and ICT
On September 8, the Ministry of Science and ICT reported a draft of the subordinate regulations of the AI Basic Act to the Presidential National AI Strategy Committee at Seoul Square. This is a follow-up measure to the AI Basic Act, which passed the National Assembly in December last year. The government plans to collect public opinions this month, announce the legislation next month, and complete the enactment by the end of the year.
In preparing these subordinate regulations, the government placed more emphasis on "promotion" rather than "regulation." By clarifying the scope of obligated entities and the criteria for regulated AI, the government has alleviated concerns about regulatory burdens and uncertainty for companies. Regulatory flexibility has been increased. If similar obligations are fulfilled under other laws, compliance with the AI Basic Act will also be recognized.
The regulations specify the criteria and procedures for designating AI clusters and similar facilities, and detail the regulatory scope, including the determination of high-impact AI, operator responsibilities, and AI impact assessments.
To establish a foundation of safety and trust, providers of generative and high-impact AI services are required to notify users in advance and must mark AI-generated outputs with watermarks or similar indicators. However, this obligation is waived if the AI is used internally by the operator or if it is clear that the service is AI-based.
Additionally, AI systems that have accumulated a certain level of training data and utilize advanced technologies, as well as systems that could significantly impact human life, physical safety, or fundamental rights, are required to ensure safety. High-impact AI operators in specific sectors such as energy, healthcare, and transportation must establish risk management organizations and implement measures for security and malfunction prevention.
The "AI Impact Assessment," which evaluates the effect of AI products and services on fundamental human rights, will be left to the discretion of operators. However, high-impact AI operators are required to make efforts to conduct such assessments.
To minimize market confusion at the initial stage of the law's implementation, the government will introduce a guidance period before imposing administrative fines. Instead of immediately issuing fines upon detecting violations, the government will encourage improvements through corrective orders and administrative guidance. The specific duration of this period has not yet been determined.
A Ministry of Science and ICT official stated, "We operated a 'Subordinate Regulation Task Force' composed of more than 80 private sector experts and collected opinions from various fields through a total of 74 meetings," adding, "We will define and implement only the minimum necessary regulations in a rational and flexible manner."
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
