Key Takeaways
- A 15-Second AI Video: $89,000 in 30 Days
- Labeling Standards Come Before Cost Per Unit
- Disclosure Responsibility Should Be Set at the Order Stage
- Count Your AI Assets and Your Disclosure Labels
A single 15-second video generated $89,000 (about 122 million KRW) in sales within 30 days of being posted. The video features a person in overalls saying, "It's on sale now, so grab it before it sells out," filmed in a style that looks like it was shot on a handheld camera. In reality, however, neither the person nor the background exists—it's an AI-generated advertisement.
According to an analysis by Vibes, an AX (enterprise AI transformation) specialist firm that collected and analyzed TikTok API data, 347,588 AI-generated videos posted on U.S. TikTok accounts in August alone generated at least $1 in sales. Of these, 76,314 videos earned between $100 and $10,000, while 514 videos brought in more than $10,000 each, and the vast majority (92.5%) were short-form clips under 30 seconds. On U.S. TikTok, there were 26,839 AI influencer accounts with 50,000 to 500,000 followers, 1,464 accounts with 500,000 to 1 million followers, and 808 accounts with more than 1 million followers. So-called "AI-native agencies"—which handle everything from ad planning to production and distribution using AI—are seeing a surge in orders from small retail brands.
Behind this growth lies the question of whether consumers can even tell that what they're seeing is AI-generated. In a survey conducted in March by British consumer group Which?, 70% of respondents were unable to accurately distinguish real footage from AI-generated video. According to The Guardian, roughly 40–60% of content associated with major brands is now AI-produced, but much of this remains undisclosed due to non-disclosure agreements (NDAs), with some AI influencer creators reportedly required by advertisers to sign such NDAs.
Regulatory efforts to improve transparency in AI advertising are also emerging. Under its AI Act, the European Union has, since August, mandated that AI-generated or AI-manipulated content be labeled so that it can be identified as such. In the United States, the Federal Trade Commission (FTC) has since 2024 prohibited fake reviews written by nonexistent people, as well as false reviews or endorsements from individuals who never actually used the product or service—rules that also apply to AI-generated reviews. That said, these regulations do not ban the use of AI models in advertising outright.
Implications for Korean Enterprise AX
For companies considering AI video advertising, it makes sense to establish labeling standards before settling on production costs. Since the EU's disclosure requirements and the FTC's review regulations apply based on the market where the ad runs, brands selling to European or U.S. consumers cannot decide on execution based solely on domestic standards. Even when production is outsourced to an agency, questions such as which party bears responsibility for disclosure, or whether the contract includes clauses about withholding AI-generation facts, are items that can and should be verified at the ordering stage.
If there's one thing to check this week, count how many of your currently running or upcoming social ad assets include AI-generated people, voices, or testimonials, and how many of those actually disclose that fact. The gap between those two numbers is the size of the risk you're currently carrying.