1.

Key Takeaways
 

• With the proliferation of products and services marketed as using artificial intelligence, there is a growing need to regulate so-called “AI washing” and false or exaggerated advertising that uses AI-generated virtual personas or deepfakes.

• To address these concerns, the Korea Fair Trade Commission released draft guidelines covering criteria for four types of unfair advertising, 16 case examples, and an eight-item self-assessment checklist, with public comments open through November 11, 2026.

• Businesses that feature AI functionality in their advertising should review how they disclose the scope, operating conditions, and limitations of their AI, confirm their substantiation, and consider submitting comments.

The Korea Fair Trade Commission (“KFTC”) announced that it will collect comments from stakeholders from October 22 to November 11, 2026 on the draft Guidelines on Labeling and Advertising Related to Artificial Intelligence (“AI”) (the “Draft Guidelines”). The Draft Guidelines set out the basic principles for businesses when labeling or advertising goods or services that use AI technology and the criteria for unfair labeling and advertising under the Act on Fair Labeling and Advertising (the “Labeling and Advertising Act”).

Businesses that use AI functionality as a key marketing element, including in electrical appliances and household goods, food, cosmetics, and household chemical products; and education, legal, and financial services, should review the Draft Guidelines closely.
 

2.

Background

Attaching the term “AI” tends to evoke sophistication, accuracy, and objectivity, and may lead consumers to perceive AI-enabled products as offering relatively superior performance and reliability. According to a consumer perception survey by the Korea Consumer Agency (November 2025), 57.9% of respondents said they would be willing to buy an AI-enabled product even if it cost more than a conventional product, paying on average about 21% more.

However, “AI washing” advertising, which claims the use of AI where none is actually used or fails to adequately disclose the scope, operating conditions, or performance limitations of the AI that is used, risks misleading consumers. Against this backdrop and with the rapid spread of false and exaggerated advertisements using AI-generated virtual personas and reviews, deepfakes, and similar content via social media, particularly in the food and pharmaceutical sectors, the government announced its Measures to Respond to False and Exaggerated Advertising Disrupting Market Order Using AI at the 7th National Policy Coordination Meeting on December 10, 2025.
 

3.

Nature and Scope of the Draft Guidelines

The Draft Guidelines are not legally binding and are intended to support voluntary compliance. The KFTC also clarified that the Draft Guidelines do "not impose new separate obligations on AI-related labeling and advertising that are more stringent than those for general products," but rather use case examples to explain how the general principles of the Labeling and Advertising Act, namely truthfulness, completeness, fairness, and substantiality, apply to AI.

Because the labeling and advertising laws do not define AI-related terms, the Draft Guidelines adopt the definitions in Article 2 of the AI Framework Act. Whether a product uses AI is determined based on the substance of its learning, inference, and judgment functions, not on how the technology is embedded or when it was trained. Products that merely respond according to preset rules, such as simple sensor or threshold-based controls, are unlikely to be regarded as AI.

In addition, the Draft Guidelines are distinct from the transparency obligations under Article 31 of the AI Framework Act (such as labeling of AI-generated output). Therefore, even if a business has complied with its transparency obligations, there may still be a violation of the Labeling and Advertising Act if the content of an advertisement is untrue or material information is omitted.
 

4.

Criteria by Type and Points for Improvement

For each of the four types of unfair advertising prohibited under Article 3(1) of the Labeling and Advertising Act, the Draft Guidelines present criteria for determining unfair AI-related labeling and advertising, together with 16 product-specific cases and before-and-after advertising examples. The key points are summarized below.
 

5.

Key Implications

Although not legally binding, the Draft Guidelines are expected to serve as a benchmark indicating the KFTC's enforcement direction under the Labeling and Advertising Act for AI-related labeling and advertising. As such, businesses should take the following points into consideration:
 

(1)

Preparing substantiation in advance: Under Article 5 of the Labeling and Advertising Act, businesses bear the burden of substantiating factual claims. Failing to submit substantiating materials within 15 days of a request under the Public Notice on the Operation of Substantiation of Labeling and Advertising may result in a cease order, administrative fines, or other measures. Even the fact of AI use should be substantiated through development, training, and operation records or license agreements. Obtaining confirmation under the AI Product and Service Confirmation Scheme (effective July 21, 2026) or "AI+" certification does not provide immunity under the Labeling and Advertising Act, nor does it substitute for substantiation of performance claims.
 

(2)

Use of third-party AI technology and professional services: Representing third-party AI technology, such as public APIs, as if it were developed in-house may give rise to deceptive labeling and advertising concerns. Businesses should review their promotional and labeling statements on the source of their technology. For professional services claiming “expert-level” capabilities, objective substantiation is expected to be a key issue.
 

(3)

AI virtual personas in food, cosmetics, pharmaceutical, and financial advertising: Under amended sector-specific laws (effective November 27, 2026), endorsements by AI-generated virtual experts in advertising for food, cosmetics, and pharmaceuticals are prohibited. Businesses in these sectors should review the methods for disclosing virtual personas under the Guidelines for Review of Labeling and Advertising Involving Endorsements and Testimonials (amended June 1, 2026), such as the phrase “[Includes virtual persona]” and a constantly displayed “Virtual persona” caption in videos, as well as the possible concurrent application of sector-specific laws.
 

(4)

Reviewing existing advertisements and submitting comments: The use of “AI” in a product name will be assessed based on whether the AI-enabled function is linked to the product's main function. In addition, the Draft Guidelines encourage businesses to voluntarily review and improve advertisements posted even before the guidelines are finalized. Stakeholders with comments on the draft may submit them to the KFTC by mail or email by November 11, 2026.