AI Leadership Requires Regulatory Overhaul for South Korea

The Urgent Need for Data Reform in Korea’s AI Development

Artificial intelligence (AI) is rapidly becoming a cornerstone of global competition. Countries that wish to maintain or enhance their technological edge must not only invest in advanced technology and infrastructure but also ensure access to large volumes of high-quality data. Data serves as both the raw material for AI development and a critical determinant of its performance. However, Korea faces significant challenges in this area, particularly when compared to the United States and China.

Korea lags behind in key areas such as technological innovation, infrastructure, and talent. Additionally, strict regulations on personal and public data have limited the availability of data for AI training. These regulatory barriers need urgent reform. Four specific issues require close attention to support Korea’s AI ambitions.

Legal Framework for Private Sector Data Use

Private companies in sectors such as electronics, telecommunications, and digital platforms often hold valuable data that could be used for AI development. However, current laws like the Personal Information Protection Act impose stringent requirements. For instance, using personal data beyond its original purpose typically requires additional consent, which can be impractical in many cases.

The “legitimate interest” clause under Article 15, Paragraph 1, Subparagraph 6 allows data use if it serves the data controller’s legitimate interests and overrides the rights of the data subject. However, the requirement for “clarity” makes this provision difficult to apply. This clause should be reconsidered or removed to enable more flexible data usage.

Moreover, a special provision should allow the use of lawfully collected personal data without explicit consent for AI development, provided it meets certain conditions. These conditions could include serving the public interest, protecting the rights of the data subject, or promoting AI innovation. Such a provision would require approval from the Personal Information Protection Commission. A bill containing these provisions has already been submitted to the National Assembly and needs swift passage.

Addressing Copyright Issues in Public Data

Publicly available data, such as news articles, can be used freely. However, copyrighted or personal data still requires permission from the copyright holder or data subject. This presents a challenge for AI systems that collect data automatically, making large-scale training difficult under current rules.

Legislation should establish clear standards for how AI systems can use copyrighted works. These standards should require proper attribution and ensure fair compensation for content creators without infringing on their rights. This would create a more balanced approach to data usage while supporting AI innovation.

Access to Public-Sector Data

Public-sector data in areas like healthcare, education, taxation, and court rulings is extensive and of high quality. However, access is often restricted due to privacy concerns or public-interest reasons. Revising individual laws to create exceptions for AI development could help, but if this proves challenging, a broader exception could be introduced through the Framework Act on Promotion of Data Industry and Data Utilization or the Basic Act on AI.

For small- and medium-sized enterprises, data trading offers an alternative source. However, this requires clearer definitions of data rights, valuation mechanisms, quality certification systems, and active marketplace development to ensure effective utilization.

Regulatory Sandboxes for AI Innovation

If immediate legal revisions are not feasible, introducing a “regulatory sandbox” for data and AI could provide a viable solution. These frameworks allow for the suspension or deferral of certain regulations for a set period, enabling experimentation and innovation. Similar models already exist in sectors like information and communications, industrial convergence, and finance.

Given the high demand for data and AI, a dedicated sandbox under the data industry act or the basic AI act would support Korea’s AI development. This approach would allow for controlled testing and refinement of AI technologies while ensuring compliance with regulatory standards.

Balancing Regulation and Innovation

The United States primarily uses ex post facto regulation for data use, while China has minimal restrictions. In contrast, Korea has adopted a European Union-style rights-based approach for personal data and AI, which may hinder its digital and AI competitiveness. The Lee Jae Myung administration, aiming to position Korea among the top three AI powers, must consider whether maintaining stricter rules than its main competitors is necessary.

In the long term, Korea should shift toward self-regulation and retroactive oversight, minimizing prior restrictions. The U.S. model, which combines punitive damages with class-action lawsuits, has proven effective in enhancing corporate accountability and improving personal data protection. Adopting similar mechanisms could offer a balanced approach, supporting both privacy safeguards and technological advancement.