An amendment to the AI Framework Act, now under committee review (referred to committee, review stage as of 8 September 2026), would expand the statutory definition of "high-impact AI." The current law already treats AI systems that materially affect hiring decisions as high-impact; the amendment would add AI systems that materially affect a worker's job assignment, task allocation, performance evaluation, and personnel management more broadly (new Article 2(4)(item 10)). This affects any AI provider whose system is used by a Korean employer or platform in these post-hiring, ongoing employment-decision contexts — not only recruitment tools.
The bill has not been enacted. No effective date or penalty figure has been set, and none should be assumed at this stage. Committee review is the relevant status to track, since a bill at this stage can be amended further, folded into a broader revision, or stall.
What this means for you
- Map your product against the expanded definition, not just hiring. If your AI system is used in Korea for job/task assignment, performance scoring, or personnel-management decisions — separate from recruitment screening — you should treat it as a candidate for future "high-impact AI" classification.
- Do not assume current thresholds carry over. No threshold, effective date, or penalty has been fixed for this amendment; nothing in the current AI Framework Act's obligations should be extended to this proposal until it is enacted.
- Track committee movement, not just the bill's existence. The practical trigger for any new compliance duty is passage and a fixed effective date, neither of which exists yet. If you already have a domestic-agent or high-impact-AI compliance program under the AI Framework Act, flag this bill for re-review at each committee stage rather than waiting for final passage to start assessing exposure.
Source: (no source URL was provided with this alert)