[보고서]Review Report on the_Defense Artificial Intelligence Basic Act_ The State That Delegates Killing to Machines

Establishing a legal framework to regulate the military use of artificial intelligence (AI) is emerging as a national priority. Amidst this trend, the proposed Defense AI Framework Act, introduced in February 2026, appears to aim at systematizing the introduction of defense AI; however, it is deeply concerning as it contains serious normative loopholes that could effectively institutionalize and accelerate the development and operation of Lethal Weapon Systems (LAWS).

In 2026, the international community stands at a decisive turning point in mandating human control over AI weapons systems through discussions at the Convention on Certain Conventional Weapons (CCW) and the related Group of Governmental Experts (GGE). The majority of countries are calling for the establishment of legally binding norms on lethal autonomous weapons, citing technical malfunctions and the ambiguity of accountability attribution. However, this bill, rather than responding to this international trend, is instead attempting to depart from international peace norms by institutionalizing a regulatory vacuum and prioritizing the convenience of research and development.

AI weapons systems must satisfy the core principles of International Humanitarian Law (IHL): notably, the principle of “Distinction” and the principle of “Proportionality.” However, fundamental doubts remain unresolved as to whether mechanical algorithms can fully replace humanitarian judgment in complex battlefield situations.

Expressions in the bill such as “human intervention” and “safety” remain abstract rhetoric lacking concrete implementation mechanisms and control measures. In particular, the “regulatory exemption” clause has significant potential for abuse as a legal basis for nullifying the strict safety and human rights standards that must necessarily apply to AI weapons systems.

Through this report, PEACEMOMO analyzes the normative crisis that the「Defense Artificial Intelligence Basic Act」(Bill) could bring about, and proposes the following comprehensive reconsideration of the current legislative direction, which is mired in technological solutionism:

  • Codification of prohibitions and restrictions on lethal autonomous weapons: Clear legal limits must be established for lethal automated weapons that escape human control.
  • Mandatory IHL compatibility assessment: A rigorous review procedure must be introduced to verify compliance with international humanitarian law from the development stage of weapons systems.
  • Establishment of independent and democratic governance: Beyond defense and industry logic, a decision-making system must be built that guarantees the substantive participation of civil society and academia, who can represent human rights and ethical perspectives.

PEACEMOMO sent this report along with an inquiry letter regarding this bill to all members of the 22nd National Assembly. Technological progress cannot precede the universal values of humanity and respect for life. We hope this report will serve as a foundation for critical understanding and discourse to revive the ethical and humanitarian responsibilities overlooked in the legislative process of defense AI, and to create institutions centered on responsibility for human dignity.

April 10, 2026

PEACEMOMO

Full Report (PDF Download)