Home > Opinion > Columns

Regulatory overhaul needed for Korea to join AI’s top three

Published: 13 Aug. 2025, 00:04

Lee Seong-yeob

The author is a professor at the Graduate School of Management of Technology at Korea University and the president of Korea Data Law and Policy Society.

Global competition over artificial intelligence is intensifying. To secure competitiveness, countries need not only advanced technology, talent and infrastructure but also large volumes of high-quality data. Data is both the raw material for AI and the main determinant of its performance.

Korea lags behind the United States and China in technology, infrastructure and talent. Strict regulations on personal and public data also hamper data use. Reforming these rules is urgent. Four issues merit close attention.

President Lee Jae Myung engages in a dialogue on artificial intelligence with Yuval Noah Harari, author of the bestseller ″Sapiens: A Brief History of Humankind″ (2011), at the National Assembly in Yeouido, Seoul, on March 22. [NEWS1]

First, private companies in fields such as electronics, telecommunications and platforms need a legal basis to use their own data for AI training. Under the Personal Information Protection Act, using personal data for AI development beyond its original collection purpose requires additional consent, which is often impractical. The “legitimate interest” clause in Article 15, Paragraph 1, Subparagraph 6 — allowing use when necessary for the data controller’s legitimate interests and clearly overriding the rights of the data subject — sets an overly strict “clarity” requirement, making it hard to apply. That provision should be deleted.

Moreover, a special provision should allow the use of lawfully collected personal data without consent for AI technology development, provided it meets certain conditions — such as serving the public interest, protecting the interests of the data subject or third parties or promoting AI innovation — and is approved by the Personal Information Protection Commission. A bill containing such a clause has been submitted to the National Assembly and should be passed promptly.

Second, public access to data such as news articles raises copyright issues. While publicly available data can be used freely, copyrighted or personal data still requires permission from the copyright holder or data subject. This is difficult to secure when AI systems collect data automatically, making large-scale training almost impossible under current rules. Legislation should set clear standards for AI use of copyrighted works, require attribution and ensure fair compensation without infringing on copyright holders’ legitimate interests.

Third, public-sector data in areas such as health care, education, taxation and court rulings is both extensive and high in quality, yet access is restricted for public-interest or privacy reasons. Revising individual laws to create exceptions for AI development would help, but if that is difficult, a blanket exception could be introduced in 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, but this requires clearer rights definitions, valuation and quality certification systems and active marketplace development.

Fourth, if immediate legal revisions are not feasible, a “regulatory sandbox” should be introduced for data and AI. Such frameworks, which suspend or defer certain regulations for a set period, already exist in information and communications, industrial convergence and finance. Given the high demand in data and AI, a dedicated sandbox under the data industry act or the basic AI act would help.

President Lee Jae Myung attends a meeting with global companies on AI cooperation at the Ulsan Exhibition and Convention Center on June 20. [JOINT PRESS CORPS]

The United States applies mainly ex post facto regulation to data use, while China has almost none. Korea, by adopting European Union-style rights-based regulation for personal data and AI, risks undermining its own digital and AI competitiveness. The Lee Jae Myung administration, which aims to make Korea one of the top three AI powers, should ask whether it must maintain stricter rules than its main rivals.

In the longer term, Korea should shift toward self-regulation and retroactive oversight, minimizing prior restrictions. The U.S. model — combining punitive damages with class-action lawsuits — has proved effective in heightening corporate accountability and improving personal data protection. Such mechanisms could offer Korea a more balanced approach, enabling both privacy safeguards and technological advancement.

This article was originally written in Korean and translated by a bilingual reporter with the help of generative AI tools. It was then edited by a native English-speaking editor. All AI-assisted translations are reviewed and refined by our newsroom staff.