인공지능의 위험에 관한 기본권 논의와 입법 대응

As artificial intelligence brings about major and minor changes in our daily lives, humanity is confronted with the complexity and autonomy of AI technologies. This is a challenge to the existing legal system that stems from the unique characteristics of A.I., which are distinct from the risks caused by the actors who use A.I.. In response, countries are beginning to regulate AI through full-fledged laws and regulations, ranging from AI ethics to AI laws. The European Union's “EU AI Act” adopts the Risiko-based management system developed in Germany in the field of environmental administrative law and requires proactive and considerate measures for the state. This shift in AI governance posture is also reflected in the NIST's AI Risk Management Framework (RMF) and the Biden administration's executive order. In Korea, the Basic Act on Artificial Intelligence was enacted in late December 2024, laying the foundation for AI regulation. This makes Korea the second country after the European Union to have a comprehensive general AI law. The Act defines high-impact artificial intelligence as “artificial intelligence systems that may significantly affect or endanger human life, physical safety, and fundamental rights,” and requires obligations to secure transparency and stability. Compared to the risk-based approach of the EU AI Act or the risk classification proposed by the US AI RMF, it is difficult to see the proportionality between the level of risk and the regulation as it contains comprehensive regulatory targets and uniform regulatory contents. It may also hinder technological innovation as the powers of regulatory authorities, such as fact-finding, are strengthened compared to other laws. Although the Korean Constitutional Court does not uniformly apply the hierarchy theory of fundamental rights, it is necessary to apply the principle of proportionality to the regulation of A.I. by differentiating the risks according to the severity of the infringement of fundamental rights and the level of regulation. On the other hand, the qualitative specificity of AI technology is autonomy, and the greatest danger to humanity from AI is the threat to human autonomous decision-making. The law and our AI regulations seem to miss the issue of violation of human dignity and self-determination, which are fundamental rights and important values of the Constitution. This should be prioritized in the future enforcement of the AI Basic Law and guidelines. The regulation of A.I. is not just a matter of jurisdiction for government regulatory bodies, but a national and time-critical issue that needs to be discussed by various experts from the private and public sectors. I hope that the normatization of A.I. will be created in the context of our technological, economic, and socio-cultural contexts to ensure sustainable A.I. innovation.