5th Market Economy Colloquium: The Direction of the Framework Act on AI

5th Market Economy Colloquium

Date and Time: 11:00 a.m., April 24, 2025, Venue: Yeollim Hall

Topic: The Direction of the AI Basic Act

Presentation:

Hea-lim Park, Senior Researcher, Center for Free Enterprise (CFE)

Discussion:

Youngyong Kim, Emeritus Professor at Chonnam National University; Yiseok Kim, Director of the Market Economy Institute;

Jaewook Ahn, Chairman of the Center for Free Enterprise (CFE);

Sung-no Choi, President of the Center for Free Enterprise (CFE); and three additional researchers from the Center for Free Enterprise (CFE)

The Direction of the AI Basic Act

Hea-lim Park, Senior Researcher, Center for Free Enterprise (CFE)

The AI Basic Act has only just taken its first legislative step. Artificial intelligence (AI), which has emerged as a core technology of the Fourth Industrial Revolution, is driving structural changes across society, the economy, and politics, going far beyond the technology itself. Technological progress is outpacing the speed of law and institutions, demanding answers not to “how should it be controlled?” but to “how should we coexist with it?” Amid this wave of change, Korea moved quickly by becoming the world’s second country to enact an AI Basic Act. However, its contents are far too rigid compared with the pace of technological change, and the business community is already raising concerns about “excessive regulation.”

We should pay attention to the American model of AI. Rather than relying on direct government-led regulation, the United States has fostered the AI market by maximizing the private sector’s capacity for technological innovation. In July 2023, major U.S. AI companies including Microsoft, Google, Meta, Amazon, and OpenAI worked with the White House to establish voluntary norms for “ensuring safe, secure, and trustworthy AI.” Although this agreement is not legally binding, it is highly significant in that it established private sector-led standards for securing the safety, transparency, and accountability of AI systems. Companies agreed to introduce risk-mitigation measures such as independent testing procedures before releasing their models and watermarking content generated by generative AI. At the same time, they also presented common standards for identifying and mitigating social risk factors such as biosecurity, cybersecurity, and algorithmic bias.

In the competitive landscape among big tech firms, a single ecosystem has naturally taken shape. Microsoft is strengthening its AI leadership through a strategic partnership with OpenAI, while Google is expanding its reach around Gemini, its own generative AI model. Apple, though often classified as a latecomer in AI, is also pursuing a differentiated path through security-centered AI chip development and a personalized model strategy. These companies have moved beyond mere technological competition and entered a stage of “norm competition,” seeking to preempt standards that guarantee AI safety, ethics, and accountability.

Korea, by contrast, is trapped in a government-led regulatory framework and moving against the current of competition. This is because it approaches AI through a preemptive blocking model based on the assumption of risk. A representative example is Article 28 of the AI Basic Act, which imposes a uniform labeling obligation on “generative AI” and places responsibility on providers. This provision does not differentiate among uses of the technology or social contexts, and instead applies the same obligation comprehensively to everything from simple summary chatbots to medical diagnosis and educational services. This is an approach that regulates the technology itself, rather than its risks.

The definition of high-risk AI is abstract, and because so much is delegated to presidential decree, it is difficult even to predict which technologies will be subject to regulation in the future. Such a legal framework leads companies to prioritize not “what can we do?” but “what might cause problems?” A structure focused on regulatory avoidance and risk minimization ultimately forces firms in the global AI war to operate in distorted ways instead of building their intrinsic competitiveness.

It has also been pointed out that a single AI service could simultaneously be subject to the Personal Information Protection Act, the Credit Information Act, and the AI User Protection Act, creating the possibility of up to “four layers of regulation.” Overlapping regulations mean that startups and companies based on new technologies may run into complex regulatory barriers before they can even commercialize their technologies. In the global race for AI dominance, where creativity and agility are essential, Korean firms are bound to face serious constraints.

To fundamentally address the limitations of the AI Basic Act, a shift in perspective is needed: rather than focusing on the technology itself, attention should be directed to the situations in which the technology is used and the impacts that may result. AI is a general purpose technology that breaks down boundaries between industries, and the level of risk is determined less by the technology itself than by its use and context. Rather than concentrating on regulating the technology, Korea should adopt a risk-based approach and establish a differentiated and flexible regulatory framework.

To build an AI ecosystem, regulatory innovation in areas such as copyright, personal data, and information security, along with AI infrastructure development and AI technology advancement, must come first. If AI uses data for information analysis purposes, exemption provisions for text and data mining (TDM) should also be discussed so that copyrighted works can be used as AI training data. Such exemptions would make it easier to secure training data for AI models and thereby contribute to enhancing corporate competitiveness.

Corporate innovation does not come from control, but from an environment in which firms can compete freely. The government should present only clear principles that do not obstruct competition, so that private sector autonomy and innovation can function properly. Now that the AI Basic Act has been introduced and the enactment of its presidential decree is approaching, it should evolve in a direction that establishes only a minimum order based on clear standards and definitions.

I hope the forthcoming enforcement decree of the AI Basic Act will become a foundation that calls forth innovation. Depending on what philosophy and standards it contains, it will determine the vitality of the AI industry. Rather than trying to catch up with the pace of technological evolution, policymakers should focus on laying the groundwork for companies to form a new AI ecosystem through institutional consistency and principles of accountability.

---

Original title: 제5회: AI 기본법의 방향

Author: Market Economy Colloquium

Date: 2025-04-24

Source: https://www.cfe.org/bbs/bbsDetail.php?cid=collo&pn=1&idx=27558