South Korea Fair Labeling and Advertising Act AI Content Disclosure: Where the Duty Actually Sits
If you don’t want to read the whole guide or have difficulty reading, you can listen to this guide instead.
Searches for south korea fair labeling and advertising act ai content disclosure assume one law does the whole job. It does not. Korea splits the work across two statutes, and mixing them up is the fastest way to misread your duties.
The duty to label AI content sits in the AI Framework Act. The Fair Labeling and Advertising Act is the enforcement tool used against misleading AI claims and undisclosed AI endorsers. This guide separates the two layers, with each rule tied to the source that states it.
AI Holic covers rules like these across our AI Governance, Ethics & Compliance section. This is a summary of published sources, not legal advice, and one layer below was still in draft when this guide was written.
Two Laws, Two Different Jobs
| Layer | Law | What it does | Status |
|---|---|---|---|
| Disclosure duty | AI Framework Act, Article 31 | Requires notice and labelling of generative AI output | In force since 22 January 2026 |
| Advertising enforcement | Act on Fair Labeling and Advertising | Punishes deceptive AI claims and undisclosed AI endorsers | In force, with 2026 guideline updates |
| Draft guidance | KFTC draft AI advertising guidelines | Explains how the advertising act applies to AI | Draft, comment period to 11 November 2026 |
| Sector bans | Sector laws, including the Cosmetics Act | Ban AI virtual-expert endorsements in set industries | From 27 November 2026 |
The Disclosure Duty Sits in the AI Framework Act
Per legal analysis published by Regpath Lab, Article 31 of Korea’s AI Framework Act requires prior notice when a service runs on high-impact or generative AI. It also requires marking of generative AI output. For virtual sound, images or video that are hard to tell from reality, users must be notified or given a label they can clearly recognise.
The Framework Act has been in force since 22 January 2026. Per the law-firm guide at decentlaw.io, breaking its labelling duty can bring a corrective order from the Minister of Science and ICT. The same guide states an administrative fine of up to 30 million KRW.
So if your question is who must label AI content in Korea, the Framework Act is your starting point. The advertising act enters when the content is an advertisement, or when an AI claim misleads buyers.
The Advertising Act Layer: AI Washing and Virtual Endorsers
The Korea Fair Trade Commission, the KFTC, enforces the Act on Fair Labeling and Advertising. Under that act, an advertisement that deceives consumers can be treated as unfair advertising, whether the deception uses AI or not.
One rule already bites in practice. Per billo.app’s summary of the KFTC’s endorsement guidelines, since 1 June 2026 AI-generated virtual people count as endorsers in Korea. A video ad must show a phrase such as “virtual person” while the figure is on screen.
Per the same source, skipping that label can count as unfair advertising under the act. The stated penalty reaches fines of up to 2 percent of related revenue. A fake expert who never discloses that they are not real is exactly the pattern this rule targets.
The KFTC Draft Guidelines of October 2026
On 22 October 2026, the KFTC released a draft document called the Guidelines on the Labeling and Advertising of Artificial Intelligence. Per reports in the Asia Business Daily and the Herald, it is open for stakeholder and expert comment until 11 November 2026.
Per those reports and the guide at aipolicydesk.com, the draft adds no new legal obligation. It applies the existing advertising act under four categories: false or exaggerated claims, deception, unfair comparisons, and disparagement. It includes 16 product-specific simulated cases and a self-assessment checklist for advertisers.
One detail matters for product marketers. Per the Asia Business Daily report, the draft says “AI” may appear in a product name only when AI technology links to the product’s primary function. Where AI plays only a supporting role, the scope of its use must be stated clearly near the product name.
The limit of the document matters just as much. Per the same coverage, the KFTC states the guidelines are not legally binding. They show how the commission reads the existing act, which is useful for planning, but the act itself remains the law.
The Sector Bans Arriving on 27 November 2026
A harder rule lands later in November. Per the Herald report and the Asia Business Daily, endorsements and testimonials by AI-generated virtual experts in food, cosmetics and pharmaceutical advertising will be prohibited from 27 November 2026. Not labelled, prohibited.
Per Regpath Lab’s legal analysis, the cosmetics part traces to an amended Cosmetics Act, Act No. 21709, promulgated on 26 May 2026. It inserts the ban into Article 13(1) of that act. The pattern is sector law doing the banning, while the Framework Act sets the disclosure floor underneath.
What This Means If You Advertise in Korea
Work through the layers in order. First, the Framework Act: if your service or content is generative AI output, plan for notice and labelling, with the clearest labels on realistic media. Second, the advertising act: never let an AI claim, comparison or endorser mislead, and label virtual endorsers while they are on screen.
Third, check your sector. A virtual expert who is merely risky in one industry is banned outright in food, cosmetics and pharmaceuticals from late November 2026. Fourth, treat the draft KFTC guidelines as a preview of enforcement thinking, including the 16 worked cases, while remembering they are not binding law.
For the wider governance picture, our guide to a unified AI governance architecture shows how companies organise these duties, and our AI governance quick wins piece covers the intake and registry habits that make compliance routine. Korea’s two-layer model is a clean example of why that organisation matters.
Frequently Asked Questions
Which Korean law requires AI content to be labelled?
The AI Framework Act, per the sources cited in this guide. Its Article 31 covers prior notice for high-impact and generative AI services, marking of generative output, and clear labelling of realistic virtual media. It has been in force since 22 January 2026.
What does the Fair Labeling and Advertising Act add?
Enforcement against deception. The KFTC uses the act against false or misleading AI claims and undisclosed endorsers. Per billo.app, an unlabelled AI virtual endorser can count as unfair advertising, with fines stated at up to 2 percent of related revenue.
Are the KFTC’s AI advertising guidelines law?
No. The October 2026 guidelines were released in draft for comment until 11 November 2026, and per the coverage the KFTC states they are not legally binding. They explain how the commission applies the existing act to AI advertising.
Can an AI expert endorse cosmetics in Korea?
Not after 27 November 2026, per the Herald and Asia Business Daily reports. Endorsements and testimonials by AI-generated virtual experts will be prohibited in food, cosmetics and pharmaceutical advertising from that date.
Conclusion
The south korea fair labeling and advertising act ai content disclosure question resolves into two layers. The Framework Act holds the labelling duty, and the advertising act punishes deception around AI claims and endorsers. Check both, then check your sector.
