A bill amending the AI Framework Act moved from committee referral to committee review on 8 September 2025. As drafted, it would add a new Article 3(3)–(5) requiring that AI development and use not produce discriminatory outcomes based on the characteristics of an individual or group, and would impose on the state a duty to manage and supervise AI-related human-rights violations and discrimination both preventively and after the fact. This is a proposed amendment, not current law — no threshold, effective date, or penalty has been set, and the existing AI Framework Act domestic-agent duty and its KRW 10 billion trigger are unaffected for now.
What this means for you
- No action required yet. The bill is at the committee-review stage; it has not passed and carries no enforceable obligation at this point.
- Track the drafting, not just the vote. If enacted, the new anti-discrimination principle and the state's supervisory duty could translate into concrete compliance expectations for AI developers and operators — for example, around bias testing, audit trails, or disclosure — even though the current text does not yet specify mechanisms or penalties.
- Distinguish this from your existing AI Framework Act exposure. This amendment targets Article 3 (basic principles), not Article 36 (domestic-agent designation). A company already assessing itself against the KRW 10 billion AI-service-revenue trigger should keep that analysis separate from this bill.
- Re-check before committee markup. Bills at this stage can be amended, merged with other proposals, or stall. Confirm status again before the next committee session rather than assuming today's text is final.
Source: