TRENKO
Guide · Korea AI Basic Act

Korea's AI Basic Act is in force. Is your system high-impact?

The Framework Act on AI applies from 22 January 2026 and reaches acts abroad that affect the Korean market. Below, free of charge: a readiness self-assessment and an obligations register for high-impact and generative AI.

Download the template Reserve your spot for an online call
§ 01 — What the Act regulates

Two categories carry almost all of the obligations.

The Act (full name: Framework Act on the Development of Artificial Intelligence and the Establishment of a Foundation of Trust) does not regulate every AI system equally. Two categories carry the weight:

Two further points decide whether the Act reaches you at all: it applies extraterritorially to acts done abroad that affect the domestic market, and operators without an address in Korea, above a threshold set by decree, must appoint a domestic representative in writing.

§ 02 — Obligation map

Which duty attaches to which category.

TriggerObligationWhat it means in practice
High-impact AIRisk managementIdentify, assess and mitigate risks across the life cycle; keep documentation
High-impact AIHuman oversightA person can monitor and intervene in the system
High-impact AIExplanationProvide, on request, the basis of the system's operation and results
High-impact AIUser protectionAdvance notice that a high-impact AI is being used
Generative AITransparencyNotify users that output is AI-generated
Generative AISynthetic markingRealistic deepfake content is clearly labelled
Foreign operatorDomestic representativeAppoint in writing above the decree threshold
Large-scale AISafetyLife-cycle risk management for high-compute systems

Category tests and thresholds are set by decree; confirm the current text before relying on them.

§ 03 — Template

Korea AI Basic Act readiness self-assessment — blank.

Copy the text below or download it as markdown. No sign-up.

KR-AIA-01 · Version 1.0 · Free Download .md ↓
# KOREA AI BASIC ACT — READINESS SELF-ASSESSMENT & OBLIGATIONS REGISTER

Document code: KR-AIA-01 · Version 1.0 · Classification: Internal
Organisation: [NAME]   Prepared by: [NAME/ROLE]   Date: ……/……/20……
Reference: Framework Act on AI (applies from 22 January 2026)

SECTION 1 — SYSTEM INVENTORY
For each AI system, record: name · purpose · provider/in-house · users ·
data categories · deployment area · Korean users (Y/N).

SECTION 2 — HIGH-IMPACT CLASSIFICATION TEST  (per system)
2.1 Is the system used in a listed area (health, energy, public services,
    biometrics, recruitment, lending, similar)?               [ ] Y  [ ] N
2.2 Can its use significantly affect life, safety or rights?   [ ] Y  [ ] N
2.3 CONCLUSION: High-impact AI?                                [ ] Y  [ ] N
    → If Y, Sections 3–4 apply.

SECTION 3 — HIGH-IMPACT OBLIGATIONS  (evidence reference for each)
[ ] Risk management across the life cycle — documented .......... ref: ……
[ ] Human oversight — who can intervene, and how ................ ref: ……
[ ] Explanation on request — basis of operation and results ..... ref: ……
[ ] User protection — advance notice of high-impact AI use ...... ref: ……
[ ] Safety measures for large-scale/high-compute systems ........ ref: ……

SECTION 4 — GENERATIVE AI TRANSPARENCY  (per generative system)
[ ] Users are told the output is AI-generated ................... ref: ……
[ ] Realistic synthetic content (deepfake) is clearly marked .... ref: ……
[ ] Placement/wording of the notice defined per medium .......... ref: ……

SECTION 5 — EXTRATERRITORIAL / DOMESTIC REPRESENTATIVE
5.1 Do you have an address in Korea?                            [ ] Y  [ ] N
5.2 If N, do you exceed the decree threshold (users/turnover)?  [ ] Y  [ ] N
5.3 If 5.2 = Y: domestic representative appointed in writing?   [ ] Y  [ ] N
    Representative: [NAME] · [CONTACT] · appointment date: ……

SECTION 6 — RECORDS
Retain: inventory, classification decisions, risk assessments, oversight
records, transparency evidence and the representative appointment.

Owner: [ROLE]   Review cycle: [PERIOD]   Next review: ……/……/20……

This template is general and does not constitute legal advice.
§ 04 — Filling it in

Three things the self-assessment does not settle for you.

  1. The high-impact test. Whether your use falls in a listed area is a legal characterisation, not a checkbox; the same model can be high-impact in one deployment and out of scope in another.
  2. Representative threshold. Whether you cross the user/turnover threshold for a domestic representative is set by decree and changes with your Korean footprint.
  3. Generative labelling. How and where the AI-generated notice must appear depends on the medium and the audience.
Note This self-assessment is a general framework. High-impact characterisation, the domestic-representative threshold and generative labelling depend on decree-level detail and on your actual Korean footprint; an incorrect high-impact call invalidates everything assessed after it. You can book a preliminary call to map your systems against the Act and set up the register.
§ 05 — Frequently asked

Questions.

When does the Korean AI Basic Act take effect?

The Framework Act on AI applies from 22 January 2026. The categories and thresholds are filled in by presidential decree, which should be checked for the operative detail.

What counts as 'high-impact AI'?

AI used in areas that may significantly affect life, safety or fundamental rights — such as healthcare, energy, public services, biometrics, recruitment and lending. The characterisation turns on the deployment, not the model alone.

Do we need a domestic representative if we have no office in Korea?

Operators without an address in Korea and above a threshold set by decree must appoint a domestic representative in writing. Whether you cross the threshold depends on your Korean user base and turnover.

Does the Act apply to acts done outside Korea?

Yes. It applies extraterritorially to acts done abroad that affect the domestic market, so a non-Korean provider serving Korean users can be in scope.

Related

Read next.

Let's map your systems against the Act.

In a twenty-minute call we run the high-impact test on your actual deployments and decide the representative and labelling questions together. If you're out of scope, we say so clearly.

Reserve your spot for an online call The call is online · the template is yours either way