On December 17, 2024, the bill for the Basic Law on the Development of Artificial Intelligence and Creation of Trust Base received the approval of the Judiciary Committee of the South Korean National Assembly. This follows the consolidation of 19 separate bills on artificial intelligence (AI) by the Science, Technology, Information, Broadcasting, and Communications Committee, which decided, on November 26, 2024, not to refer the 19 bills to the plenary session of the National Assembly.

Definitions

The bill defines AI as 'an electronic implementation of intellectual capabilities that humans possess, such as learning, reasoning, perception, judgment, and language comprehension.'

Notably, the bill considers 'high-impact AI' as 'an AI system that may have significant effects on or pose risks to human life, bodily safety, or fundamental rights.' This includes AI used in areas such as energy production, drinking water production, the provision and use of healthcare services, and biometric information, where there is a significant impact on persons rights and obligations such as hiring or loan screening, and transportation.

The bill also provides definitions for 'generative AI,' 'AI business operator,' 'AI system,' and 'AI technology,' among others.

Principles

The bill sets out basic principles for AI, namely that:

• AI technology and the AI industry shall be developed to enhance safety and reliability;
• state and local governments must respect the creativity of AI business operators to create an environment conducive to safe AI use;
• state and local governments must devise policies to ensure that AI stably changes society; and
• individuals shall be provided with a clear and meaningful explanation of the reasons and principles behind the results produced by AI.

Obligations

Specifically, the bill details transparency requirements, including:

• business operators providing AI products or services using high-impact AI or generative AI must notify users in advance;
• if providing generative AI products or services, business operators must indicate the results generated by generative AI; and
• if the results are difficult to distinguish from reality, AI business operators must clearly inform users of results created by an AI system.

Regarding AI safety, business operators whose AI system exceeds a threshold of cumulative computational usage for training as determined by a future Presidential Decree, must:

• identify, assess, and mitigate risks throughout the AI lifecycle; and
• establish a risk management system to monitor and respond to AI-related safety accidents.

Business operators must review in advance whether AI systems constitute high-impact AI and confirm such qualification with the Ministry of Science and ICT. Business operators that do provide high-impact AI services must, among other things:

• develop and operate a risk management plan;
• implement measures to explain the final AI output and criteria used to derive it;
• develop and operate user protection measures;
• ensure human oversight of high-impact AI; and
• prepare and retain documentation on measures taken to ensure safety and reliability.

High-impact AI business operators must also conduct an impact assessment in advance of the effect on fundamental rights.

In addition, AI business operators that do not have an address or business office in South Korea and that meet criteria to be determined by a Presidential Decree must designate a domestic representative and report this to the Ministry of Science and ICT. The domestic representative is responsible for many of the obligations under the bill, including submitting results on whether an AI reaches the definition of high-impact AI.

Enforcement

The bill details fines of up to KRW 30 million (approx. $20,872) for violations of specific provisions, alongside the potential imprisonment of individuals for up to three years in case of the disclosure of confidential information learned in the course of duty.

The Ministry of Science and ICT is responsible for developing implementation plans under the bill, clarifying respective obligations and criteria. More specifically, the Ministry of Science and ICT will develop a Basic AI Plan every three years to promote AI and direct AI policy. The Ministry of Science and ICT may also designate an AI Policy Center and AI Safety Research Institute, with the responsibilities of the Policy Center and Research Institute detailed under the bill.

Likewise, the bill mentions the potential development and publication of AI Ethics Publication by the Ministry of Science and ICT.

Future guidance to be published pursuant to the bill includes measures on AI training data, support measures for small and medium-sized enterprises (SMEs), alongside measures to establish AI testbeds and promote AI data centers.

You can read the bill and track its progress, only available in Korean, here.