South Korea's Fair Trade Commission seeks comment on draft AI advertising guideline
In short
South Korea's Fair Trade Commission said it will take opinions from 22 October to 11 November on a draft guideline for AI-related labeling and advertising. The draft targets "AI washing", meaning products advertised as using AI when they do not, and false or exaggerated ads using deepfakes or AI-generated people. The guideline is not legally binding.
Read the full story 2 min read
South Korea's Fair Trade Commission said it will take opinions from 22 October to 11 November on a draft "AI-related labeling and advertising guideline". The commission said it is seeking views from businesses, consumers, experts and other stakeholders. It plans to finalise and distribute the guideline after the comment period and consultation with related ministries. [ 1 , 2 , 3 ]
According to Newsis and Yonhap, the guideline aims to prevent "AI washing", meaning labeling or advertising a product as using AI when it does not. It also covers cases where AI is used but its scope, operating conditions or limits are not made clear. Yonhap added that it also addresses consumer harm from false or exaggerated ads using AI-generated people, reviews and deepfakes. [ 1 , 2 ]
The draft sets out basic principles and criteria under the labeling and advertising law for four types of improper ads: false or exaggerated, deceptive, unfair comparison and disparaging. Newsis and Asia Economy said it includes 16 hypothetical cases, with before-and-after ad examples. Newsis listed sectors with a higher chance of consumer harm: electrical and household goods, food, cosmetics and household chemical products, and education, legal and financial services. [ 1 , 2 , 3 ]
Yonhap reported that promoting a product as an "AI lawyer", "AI electric hot water mat" or "AI customized shampoo" when it has no AI technology, or only a simple or auxiliary level, can count as a false or exaggerated ad. Hiding key matters such as operating conditions while using AI can be treated as deceptive. Comparisons such as "X times better than other companies" that omit the target, the standard or laboratory conditions are improper, and belittling existing technology as outdated or fake while presenting a company's own AI as perfect can be read as disparaging. [ 2 ]
Newsis said the draft also contains an eight-item self-check list for businesses, consumer notes and frequently asked questions. It said the guideline is not legally binding and adds no new duties; it explains the existing general principles of truthfulness, completeness, fairness and substantiation through cases. [ 1 ]
Newsis and Asia Economy cited a Korea Consumer Agency survey from last November. In it, 57.9% of respondents said they would pay more for products with AI technology, by an average of about 21%. Asia Economy separately reported that the draft allows "AI" in a product name only when the technology is tied to the main function, and that AI virtual experts' recommendations in food, cosmetics and medicine ads are banned under amended laws. [ 1 , 3 ]
Yonhap reported that the commission prepared the draft with the Korea Consumer Agency through expert forums and business meetings. After the comment period it plans to revise the guideline periodically. Newsis said comments can be sent by post or email. [ 1 , 2 ]
Why it matters
The draft sets out how the commission reads existing advertising law for AI claims, with examples for businesses. Newsis reported it adds no new obligations and has no legal force, so its effect depends on the final text after the comment period.
Key facts
- The Fair Trade Commission said it will collect opinions on a draft "AI-related labeling and advertising guideline" from 22 October to 11 November. [ 1 , 2 , 3 ]
- The draft addresses "AI washing", advertising products as using AI when they do not, as well as ads that hide the scope, operating conditions or limits of AI features. [ 1 , 2 ]
- It sets criteria for four types of ads banned by the labeling and advertising law: false or exaggerated, deceptive, unfair comparison and disparaging. [ 1 , 2 , 3 ]
- The draft includes 16 hypothetical cases with before-and-after ad examples, according to Newsis and Asia Economy. [ 1 , 3 ]
- Newsis reported that the guideline is not legally binding and adds no new obligations to AI-related advertising. [ 1 ]
- The commission plans to finalise and distribute the guideline after the comment period and consultation with related ministries. [ 1 , 2 , 3 ]
Confirmed by several sources
- The Fair Trade Commission is taking opinions on the draft AI labeling and advertising guideline from 22 October to 11 November. [ 1 , 2 , 3 ]
- The guideline covers four types of improper ads: false or exaggerated, deceptive, unfair comparison and disparaging. [ 1 , 2 , 3 ]
- After the comment period and consultation with related ministries, the commission plans to finalise and distribute the guideline, and to update it periodically (Newsis and Yonhap). [ 1 , 2 ]
Still unclear
- Whether AI virtual experts' recommendations in food, cosmetics and medicine ads are banned under amended laws, as Asia Economy reported. Only Asia Economy states this; the other documents do not mention it.
- Rules on using "AI" in product names, which Asia Economy said the draft sets out. Only Asia Economy describes this; the other documents do not mention it.