1. Discussion of Fundamental Rights Regarding the Risks of Artificial Intelligence and Legislative Response, Lee Hee-ok, 2025.
2. Research on the Artificial Intelligence Basic Act and Ethics in the AI Era, Kim Gyeong-dong, 2025.

 Korea enacted the Artificial Intelligence Basic Act in late 2024, becoming the world's second country after the European Union to have a comprehensive AI general law. However, Lee Hee-ok and Kim Gyeong-dong's papers criticize that this law, in its rush to respond technically and industrially, is missing the most essential values the law must protect. The authors' core argument is that Korea's AI Basic Act may have addressed physical safety of citizens' lives and bodies, but has left human dignity and autonomy — spiritual values — in a legislative blind spot.

Discussion of Fundamental Rights Regarding the Risks of Artificial Intelligence and Legislative Response, Lee Hee-ok, 2025.


Redefining Risk: From Error and Discrimination to 'Erosion of Autonomy'

While existing AI risk discussions focused on technical errors or legal disputes such as data bias, copyright infringement, and hallucinations, this paper shifts the weight center of risk to 'the hierarchy of constitutional fundamental rights.' The author distinguishes AI risks into two levels. The first is 'aggravated legal risk' addressable by the existing legal system, and the second is 'critical risk' arising from the qualitative characteristic of AI — autonomy.

The true fear the author focuses on is the latter. A situation where AI goes beyond assisting human decision-making to completely replacing human judgment or humans becoming subordinated to AI decisions. This is not simply a technological side effect but a threat that fundamentally shakes the 'right to self-determination' and 'human dignity' guaranteed by the Constitution. However, current discussions are buried only in visible problems like deepfakes and copyright, remaining silent about the structural danger of AI eroding human autonomy. The author categorically states that this "misses the problem of violating constitutional values of human dignity and self-determination."

Korea's AI Law That Became a 'Blunt Blade' Losing Precision

The paper evaluates Korea's Artificial Intelligence Basic Act as a 'hybrid' clumsily mixing the EU and U.S. systems. The EU's 'AI Act' thoroughly adheres to the 'principle of proportionality' that adjusts regulatory intensity by subdividing risk into 4 levels (prohibition-high risk-limited-minimal). In contrast, Korea's bill has piled everything from threats to life and body to general fundamental rights impact into one giant basket called 'high-impact artificial intelligence.'

This uniform approach has blunted the regulatory blade. Because 'dignity infringement' risk that should be treated more importantly from constitutional values and relatively light 'service risk' are treated on the same level. This inherits a structural contradiction that will inevitably result in either over-regulation or under-regulation. The author worries that this legislative attitude lacks proportionality between risk levels and regulations and will ultimately result in neither protecting citizens' safety nor inhibiting corporate innovation. In particular, the metaphor that trying to transplant Europe's strong regulation as-is without consideration for Korean companies' reality would give companies the pain of an unsolvable 'cube puzzle' captures field concerns.

Absence of Constitutional Decision on Human Dignity

The most pointed criticism in the paper is that there is no substantive will to protect 'human dignity' and 'right to self-determination' anywhere in Korea's bill. The EU defines AI that manipulates human subconscious or evaluates humans by assigning social scores (Social Scoring) as 'prohibited risk' and fundamentally blocks it. This reflects Europe's firm philosophy that human dignity precedes technology.

In contrast, Korea's law specifies safety of 'body and life' but remains silent on situations where mental autonomy is infringed. The author laments that this shows the cross-section of Korea's legal system failing to recognize constitutional human dignity as substantive norms. When AI tries to replace human uniqueness, our law does not have a shield to protect humans. The author argues that this part must be necessarily supplemented even in future enforcement regulations or guidelines, and that resolute legislative decision is needed regarding AI that threatens humans' autonomous decision-making.

Ambiguous Labeling Obligation Violating the Principle of Clarity

A part particularly noteworthy for creators and the graphics industry is the ambiguity of 'labeling obligations.' The bill stipulates that AI-generated products must be labeled, but in a way that "users can clearly recognize" while simultaneously "not inhibiting the enjoyment of artistic and creative expressions." The author points out that this provision is excessively abstract, it is impossible to know how far business operators' discretion is permitted, and there is a risk of violating the constitutional 'principle of clarity.' This obligation thrown without concrete guidelines between users' right to know and the aesthetic value of works will inevitably cause great confusion in the field.

Regulatory Redesign Through 'Fundamental Rights Impact Assessment'

The alternative the author presents is the introduction of 'fundamental rights impact assessment.' Going beyond technology impact assessment that simply looks at technology's impact on society, the constitutional fundamental rights impact of AI must be empirically analyzed. Through this, the gravity of risks must be distinguished, and truly 'qualitative risks' that must be prohibited and 'legal risks' that may be permitted but managed must be separated.

 The paper concludes that Korea's AI regulation must escape government-led rush legislation and advance toward sustainable governance centered on 'human dignity' through intense discussion by civilian and government experts. This is not a technological problem but a constitutional decision on what values our society will prioritize.



Research on the Artificial Intelligence Basic Act and Ethics in the AI Era, Kim Gyeong-dong, 2025.

The Law That Deals With Technology, But Missed Technology

This paper persistently digs into not what Korea's 'Artificial Intelligence Basic Act' passed in late 2024 sought to regulate but what it failed to regulate. The author sees that even though this law presents the pretext of 'fostering the AI industry and creating a trust base,' it seriously lacks the minimum requirement a law must have — legal clarity. In particular, the fact that core concepts that should manage AI risks are presented in excessively abstract and comprehensive ways, with responsibility and regulation standards effectively deferred to enforcement decrees and presidential decrees rather than legislation, is taken as a problem. As a result, the evaluation is reached that the basic law remains as a directional declaration and is difficult to operate as norms where actual responsibility can be held.

The Politics of Language Called 'High-Impact' Not 'High-Risk'

The core legal criticism this paper raises lies in the choice of terminology used in the bill. Unlike the EU's AI Act, which clearly uses the term 'High-risk' to specify regulatory targets, Korea's basic law adopted the neutral and ambiguous term 'High-impact.' The author diagnoses that even if this was a political choice originating from intent to mitigate the industry's negative reactions, it has seriously damaged legal stability as a result. The question "What exactly is high-impact?" is not answered clearly by the law, deferring all specific criteria and responsibility attribution to enforcement decrees. This ultimately destroys the consistency of regulatory application and results in the disappearance of accountable subjects before actual risks AI will cause — a pointed criticism.

 The Problem Is Not Algorithms but Human Desire

This paper reduces and interprets the cause of ethical problems not to technology but to 'human nature.' From the author's perspective, AI algorithms are not value-neutral mathematical products but mirrors of desire reflecting designers' intentions and profit-seeking (utility). That is, ethical problems AI causes are not machine malfunctions but phenomena where human moral hazard and ethics collapse are amplified through technology as medium. Therefore the paper asks not "Why is AI dangerous?" but "How does human desire outpace law and ethics?" 

Technology Autonomously, Responsibility on Humans

 It argues that since it is both impossible and could impede industrial competitiveness to rashly constrain technology itself through law, open and autonomous regulation should be applied to AI generation and distribution. Instead, the regulatory blade should be aimed not at technology itself but at the 'humans (actors)' who design and own those algorithms. That is, a 'dual approach' is needed that guarantees autonomy of technology development while strictly holding designers and owners accountable for results arising from it. This means a transition from technology-centered regulation to actor-centered regulation, and is presented as a realistic alternative capable of utilizing AI's potential while controlling moral hazard.

Aim at Subjects, Not Tools 

In conclusion, this paper delivers the message that law should accurately aim at humans who wield technology rather than rashly judging or defining technology. Criticizing the legislative laziness hidden behind the ambiguous word 'high-impact' and defining the essence of AI ethics not as technology but as human desire carries significant implications. Law and ethics in the AI era depend not on obsessing over technical control but ultimately on how to institutionalize responsibility behind technology — this is the suggestion this research leaves.