South Korea plans to require labels on AI-generated content, part of a government effort to counter fake advertisements and deceptive synthetic media circulating online.

The measure targets a growing problem for platforms, advertisers, and regulators alike: synthetic audio, video, and images produced by generative AI tools are now cheap and fast enough to produce at scale, and many consumers cannot distinguish them from authentic material. Fake advertisements built on this technology have become a particular concern, because they lend false credibility to products and financial schemes.

Under the planned rules, content created or substantially altered by artificial intelligence would have to carry a disclosure identifying it as machine-generated. The goal is straightforward: give consumers a clear signal about what they are looking at before they buy, invest, or share.

Why the regulator is acting now

Generative AI has collapsed the cost of producing convincing fake media. What once required a production studio now takes minutes on a consumer laptop or a phone. That shift has put realistic fake endorsements, fabricated product demonstrations, and cloned voices within reach of ordinary scammers, not just organized fraud rings.

Fake advertisements are among the most damaging applications. A synthetic video of a well-known figure promoting an investment product or a health claim can spread rapidly on social platforms before anyone verifies it. By the time the content is taken down, the damage — in purchases, subscriptions, or outright fraud — is often already done.

Mandatory labeling aims to break that cycle at the point of consumption. If a viewer knows a video is AI-generated, the persuasive weight of a fake endorsement drops sharply.

What the rules would require

The core obligation falls on those who create or distribute AI-generated material: mark it as synthetic. Disclosure requirements of this type typically apply across formats — video, audio, images, and text — because fake advertising campaigns mix all of them.

For platforms, that means detection and labeling mechanisms become part of compliance, not an optional trust-and-safety feature. For advertisers and agencies, it means synthetic content used in campaigns must be declared rather than passed off as authentic footage.

Enforcement questions remain

As with any labeling regime, enforcement is the hard part. Regulators must decide who bears primary responsibility — the creator, the publisher, or the platform — and what penalties apply when labels are missing or deliberately stripped.

Detection adds another layer of difficulty. Watermarking embedded at generation time can be removed or degraded through re-encoding, cropping, or filters. Post-hoc detection tools exist but remain in an arms race with generation models that keep improving.

Korea is not alone in confronting this. Disclosure mandates for synthetic media have been moving onto legislative agendas in multiple jurisdictions, and platforms have rolled out their own AI-content labels. A national mandate, however, sets a uniform legal floor rather than a patchwork of voluntary policies.

What it means for the market

For operators in advertising, media, and e-commerce doing business in Korea, the direction of travel is clear. Synthetic content is not being banned — it is being made transparent. Campaigns that use generative AI will need labeling workflows baked in from production through distribution.

For consumers, the payoff is a simpler decision at the moment of exposure: a label that says, in effect, treat this differently. Whether that label arrives reliably on every fake ad is the question the enforcement details will answer.

The mandate signals that Seoul treats AI-generated deception as a consumer-protection issue first. Advertisers and platforms operating in the Korean market should expect the labeling obligation to apply to them and prepare compliance processes before the rules take effect.