From this year, AI operators will face greater responsibility for risk management. Support for supplying advanced GPUs to industry, academia and research will begin in earnest.

On Jan. 2, the government and the National Assembly said the IT and broadcasting and telecommunications industries are expected to gain momentum this year in AI governance and in securing globally competitive, homegrown AI models. Sanctions on malicious distribution of information will also be strengthened, including the introduction of punitive damages for distributing false, manipulated information.

AI output labeling to become mandatory; risk management responsibility to rise

The Framework Act on the Development of Artificial Intelligence and Establishing a Foundation of Trust, known as the AI Basic Act, will take effect on Jan. 22. It sets specific AI governance to secure safety as the government pushes to become one of the world’s three leading AI powers.

A key part of the AI Basic Act is that it includes obligations to ensure AI transparency and safety, standards for high-impact AI and the responsibilities of operators. The Ministry of Science and ICT said the goal is to promote AI through the minimum necessary regulation.

Under the enforcement decree of the AI Basic Act, operators must provide prior notice or markings such as a watermark when they provide products or services using high-impact AI or generative AI. The risk management and oversight responsibilities of AI operators have increased, including requiring operators to identify risk factors and draw up emergency response plans for AI systems that apply advanced technology among AI systems of at least 10 to the 26th power flops.

The government will operate a guidance period for administrative fines in the early stage to minimize confusion among companies. It will delay for at least 1 year the imposition of administrative fines of up to 30 million won for violations of the enforcement decree of the AI Basic Act. It has virtually the same effect as a regulatory grace period.

Use of advanced GPUs will also begin in earnest. The Ministry of Science and ICT secured about 13,000 GPUs through a supplementary budget last year. From February, it will begin distributing 10,000 units for government use to industry, academia and research, and national-level AI projects.

The ministry began accepting project applications from industry, academia and research through an online platform in December. It is also identifying national projects to drive national AI innovation such as sector-by-sector AX through demand surveys by relevant ministries.

In the second half of this year, the government will secure a homegrown AI model ranked in the global top 10 through the "independent AI foundation model" project. Five elite teams are currently developing an independent AI foundation model: Naver Cloud, Upstage, SK Telecom, NC AI and the LG AI Research Institute.

The government will gradually release these models and provide them in open-source form to encourage wider ecosystem adoption. It plans to eliminate 1 team through the first stage evaluation this month and advance 4 teams to the next stage.

Up to 5 times damages for spreading false, manipulated information

From the second half of this year, those who maliciously spread false, manipulated information must compensate up to 5 times the damages caused by distributing that information.

In December, the Cabinet approved a promulgation bill for the Act on Promotion of Information and Communications Network Utilization and Information Protection. The bill is known as the "Act to Eradicate False and Manipulated Information".

Under the amendment, if a person is both a "poster prescribed by presidential decree" and "a person whose business is conveying facts or opinions," and the distribution of false, manipulated information satisfies intent, purpose and infringement of legally protected interests, a court may set increased damages within a range of up to 5 times the damages.

If information that has been deemed illegal or false and manipulated is repeatedly distributed on information and communications networks after a criminal guilty verdict, damages ruling or corrective report ruling becomes final, the Korea Communications Commission can impose an administrative penalty of up to 1 billion won.

The amendment to the Network Act has drawn criticism mainly from civil society that it will shrink press freedom and freedom of expression. In particular, the U.S. government has expressed concern, raising the possibility it could escalate into a diplomatic conflict.

In a spokesperson’s response sent on Dec. 31 to a query asking for the State Department’s position on the law, the U.S. State Department said, "The United States has serious concerns that the South Korean government has approved amendments to the Network Act that negatively impact U.S.-based online platform businesses and weaken freedom of expression."

The government and the ruling party that led the passage of the law say the increased damages will exclude information aimed at the public interest, including information related to acts that harm the public interest under the Act on the Protection of Public Interest Whistleblowers, information on acts prohibited under the Kim Young-ran Act, and information of comparable public interest.