A proposed amendment to the AI Framework Act (in force since 22 January 2026) would add explicit anti-gender-bias language to the Act's basic principles and require impact assessments for high-impact AI to evaluate whether gender bias or discrimination occurs, with remedial measures required based on the results. The bill would amend Article 3(3), add a new Article 27(1)(4), and amend Article 35. It affects businesses that develop or operate high-impact AI systems in Korea. As of 2026-09-08, the bill has moved from committee referral to committee review — it has not been enacted, and no effective date, threshold, or penalty has been set.
What this means for you
- No action required yet. This is a pending bill under committee review, not current law. The existing AI Framework Act's ethics principles and high-impact AI impact-assessment duties remain as they are today.
- If you develop or operate high-impact AI systems reaching the Korean market, take note that the scope of impact assessments may expand to explicitly cover gender bias and discrimination, with a follow-on duty to implement corrective measures based on findings.
- Track committee progress. Bills can stall, be revised, or be folded into a broader revision before passage — the scope and any threshold or penalty could change before enactment.
- Revisit your impact-assessment methodology if you already conduct one, so you are not caught unprepared if a bias-evaluation requirement is formally added.
Source: