AI Ad Disclosure Laws by Country: EU, US, South Korea 2026

Table of Contents:

• What Countries Require AI Disclosure in Advertising in 2026?
• Enforcement Timeline at a Glance
• Jurisdiction Deep-Dives
• Side-by-Side Comparison: All Five Jurisdictions
• Where the IAB Framework Fits
• Frequently Asked Questions

What Countries Require AI Disclosure in Advertising in 2026?

If you are wondering what countries require AI disclosure in advertising, the list is growing rapidly in 2026. Until recently, "AI advertising disclosure" was a theoretical concern. Today, it is an operational requirement. The wave of legislation that compliance teams have been tracking is now breaking onshore — South Korea's rules are live, New York's clock is counting down to June, and California's SB 942 follows two months later. Meanwhile, the EU AI Act is building to full enforcement in 2027, and the US Federal Trade Commission has already used its existing deceptive-practices authority to pursue undisclosed AI content.

For global agencies managing multi-market campaigns, the risk is not that any one law is hard to follow in isolation — it is that the five frameworks are meaningfully different from each other. A disclosure label that satisfies California's human-readable requirement may still fail South Korea's placement rules. A consent workflow designed for New York's synthetic performer standard will not cover every touchpoint the EU cares about.

This complete 2026 guide explains what each framework requires, compares them side-by-side, and maps how the IAB AI Transparency and Disclosure Framework fits on top of all five.

Enforcement Timeline at a Glance

For agencies distributing ad creative globally, here is the immediate enforcement timeline:

• Live now: South Korea Telecoms Act
• Already enforced: US FTC Endorsement Guidelines
• June 2026: NY Synthetic Performer Law
• Aug 1, 2026: California SB 942
• Aug 2026 / 2027: EU AI Act (phased)

Jurisdiction Deep-Dives

🇪🇺 European Union — EU AI Act (Phased Rollout)

Key Definition: The EU AI Act is a comprehensive regulatory framework that governs the sale and use of artificial intelligence in the EU, mandating that providers and operators of AI systems label synthetic media so consumers know it is machine-generated.

Core obligation: Article 50 of the EU AI Act requires providers of AI systems that generate synthetic media — including images, video, audio, and text that a "reasonable person" would mistake for authentic human-created content — to ensure that output is marked as AI-generated in a machine-readable format. Operators (i.e., agencies and advertisers deploying those systems) must make that information available to end users in a "clear and distinguishable manner."

Advertising-specific trigger: Any AI-generated or AI-altered image, video, or audio in a paid commercial context is covered. AI-assisted copy (where a human substantially edits AI output) sits in a grey zone, but the IAB Framework's materiality test provides a useful bright-line guide for practical compliance decisions.

Enforcement timeline: GPAI model transparency requirements took effect in August 2025. Obligations for deployers — the agencies and brands configuring AI for their campaigns — take full effect in August 2026. Content labelling for all synthetic media reaches full force in 2027.

Penalty exposure: Fine: Up to €15M or 3% global revenue for violations by operators, whichever is higher. National competent authorities in each member state enforce, meaning enforcement intensity will vary across the 27-country bloc in early years.

Key nuance: The EU Act includes a narrow exception for creative, satirical, or parody content — but it still requires an "appropriate disclosure" even in those cases, unless the disclosure would "interfere with the display or enjoyment" of the content. Do not read this as a blanket art-direction exemption.

🇺🇸 United States (Federal) — FTC Guidelines (Enforceable Now)

Key Definition: The FTC Endorsement Guidelines are federal rules enforcing truth-in-advertising that classify undisclosed or deceptive AI-generated endorsements, reviews, or testimonials as unfair or deceptive acts under Section 5 of the FTC Act.

Core obligation: The FTC does not have a single "AI disclosure law." Instead, it applies its long-standing Section 5 authority over unfair or deceptive acts and practices, combined with its 2023 revised Guides Concerning the Use of Endorsements and Testimonials, to AI-generated content. The FTC's position is clear: if AI is used to generate or fabricate an endorsement, review, or testimonial that a consumer might believe is from a real person, that is a deceptive practice. Disclosure is required.

Advertising-specific trigger: The trigger is consumer deception, not AI use per se. A visually obvious AI illustration does not require a label under FTC theory; a photorealistic AI spokesperson who appears to be a real customer does. The IAB's "materiality" test — would a reasonable consumer act differently if they knew this content was AI-generated? — closely mirrors FTC logic.

Enforcement timeline: No future date to wait for. The FTC has issued warning letters and initiated investigations into AI-generated testimonials. The agency is actively watching.

Penalty exposure: Fine: Up to $51,744/day per violation for knowing violations of FTC rules. More practically, consent decrees and remediation orders can impose sweeping operational changes on advertisers.

Key nuance: The FTC's guidance explicitly calls out negative reviews suppressed by AI as a deceptive practice — disclosure is not just about labelling generated content, it's also about ensuring AI curation doesn't create a misleading picture of consumer sentiment.

🗽 New York State — Synthetic Performer Law (Effective June 2026)

Key Definition: The NY Synthetic Performer Law is a state civil rights statute that prohibits the commercial use of a living or deceased person's voice, face, or performance replicated by AI without their explicit written consent.

Core obligation: New York's statute targets a specific and high-value use case: the AI-generated replication of a real person's voice, face, likeness, or performance in a commercial advertisement. Any advertiser or agency doing this without explicit written consent from the living individual — or from the estate of a deceased performer — is in violation. On-screen or audible disclosure is also required.

Advertising-specific trigger: The law is narrowly scoped to commercial advertising. It does not apply to news, editorial, satire, or political speech. If your ad features an AI-generated voice that sounds like a recognisable celebrity, a reanimated historical figure, or a synthetic replica of a real person — and it runs in New York — you need consent documentation before June 2026.

Enforcement timeline: Effective June 2026. New York's Attorney General has enforcement authority, and private right of action is available to affected individuals or their estates.

Penalty exposure: Fine: Up to $1,000 to $5,000 per violation. Given that a single national ad campaign can generate thousands of "impressions" in New York, this can compound rapidly. Class actions are a real risk given the private right of action.

Key nuance: The consent requirement extends to deceased performers for 40 years post-death, and the estate right is descendible. Agencies repurposing archival audio or building AI voice models from licensed recordings of deceased artists need to verify estate consent is in scope for the NY statute specifically.

🌴 California — SB 942 AI Transparency Act (Effective Aug 1, 2026)

Key Definition: California SB 942 (The AI Transparency Act) is a state-level AI regulation that requires large AI providers to embed detectable provenance signals (C2PA) in AI-generated content and offer a free AI detection tool to users.

Core obligation: California SB 942 targets the AI providers (the companies making the tools) more than the advertisers themselves — but its downstream effects are significant for ad teams. Any covered AI provider must: (1) embed detectable watermarks or provenance metadata in AI-generated content, (2) offer a free tool allowing users to detect AI-generated content, and (3) retain provenance records for five years. Advertisers distributing AI creative through covered platforms inherit the labelling obligations.

Advertising-specific trigger: Any AI-generated image, video, or audio distributed commercially in California through a covered platform. "Covered platform" means a provider deploying a generative AI system with over one million monthly California users — which includes virtually every major ad-tech platform.

Enforcement timeline: August 1, 2026. The California Attorney General enforces.

Penalty exposure: Fine: Up to $5,000 per unlabelled piece of content. Because SB 942 also mandates machine-readable provenance (C2PA-compatible), the penalty clock starts when content is distributed without valid metadata — not just when a consumer complains.

Key nuance: SB 942 is the only US state law that explicitly requires machine-readable provenance, not just a human-readable label. This makes it the US law most technically aligned with the IAB Framework's C2PA assertion requirements. If you build a C2PA-compliant workflow for California, you have a head start on IAB compliance for every other market.

🇰🇷 South Korea — Telecoms Business Act (Live Now)

Key Definition: South Korea’s Telecoms Business Act is a national telecommunications framework that mandates broadcasters and platform operators to attach visible disclosure labels to AI-generated or AI-altered advertising content throughout its entire duration.

Core obligation: South Korea moved faster than any Western regulator. Amendments to the Telecoms Business Act — effective early 2026 — require broadcasters, platform operators, and digital service providers to attach clear, visible disclosure labels to AI-generated or AI-materially-altered advertising content. The disclosure must appear throughout the duration of the ad, not only at the beginning.

Advertising-specific trigger: All AI-generated or AI-altered commercial content distributed via broadcast or digital channels in South Korea. Platform operators (not just advertisers) carry the compliance duty, which means demand-side partners serving Korean inventory are already in scope.

Enforcement timeline: Already in effect. The Korea Communications Commission (KCC) oversees enforcement, and fines have already been issued to broadcasters in test cases.

Penalty exposure: Fine: Up to KRW 30M (~$22K USD) per violation, with potential licence suspension for persistent violations. The platform-operator liability model means media agencies buying South Korean inventory face indirect exposure if their supply-side partners are fined and seek indemnification.

Key nuance: South Korea's requirement that the disclosure label remain visible throughout the entire ad — not just in a brief end-card disclaimer — is stricter than any US rule and stricter than current EU guidance. If your global creative template uses a 3-second end-card disclosure, it will fail the Korean standard.

Side-by-Side Comparison: All Five Jurisdictions

The table below is designed for structured comparison. Use this alongside our compliance check tools designed for ad-tech teams running cross-border campaigns.

Note: This table reflects the best available interpretation of each framework as of March 2026. It is not legal advice. Always consult qualified counsel for jurisdiction-specific compliance decisions.

Where the IAB Framework Fits

The IAB AI Transparency and Disclosure Framework is not a law — it is the advertising industry's self-regulatory response that sits on top of all five legal frameworks above. Think of it as the compliance floor that is consistent across markets, even when the laws themselves are not.

The Framework introduces three tools that are relevant to multi-jurisdiction compliance:

1. The Materiality Test. The IAB's central question — "would a reasonable consumer make a different decision if they knew this content was AI-generated?" — provides a single bright-line that approximates the trigger conditions in all five jurisdictions.
2. C2PA Assertions. The Framework mandates embedding IPTC digitalSourceType vocabulary assertions in any AI-generated creative. This directly satisfies the machine-readable requirement in the EU AI Act and California SB 942.
3. Tiered Disclosure Labels. The IAB specifies label language, placement, and duration standards. Building your creative templates to the IAB's Tier 2 label standard effectively makes them South Korea-compliant by default.

Conclusion: The Bottom Line for Global Campaigns

The patchwork nature of AI advertising disclosure law in 2026 creates real operational complexity — but it also creates a clear compliance architecture if you know where to look. South Korea demands action immediately. New York's June deadline is urgent for any campaign using AI-generated voices or likenesses. California's August deadline is technically demanding because of C2PA. The EU's full force arrives in 2027, but proactive C2PA implementation now is the right infrastructure investment.

The practical playbook: build your creative workflows to the IAB Framework standard, implement C2PA metadata at the tooling level, adopt the IAB's Tier 2 label placement rules as your global default, and layer the New York consent documentation workflow on top for any campaign using identifiable voice or likeness. That combination covers all five frameworks currently in force or coming into force in 2026.

Frequently Asked Questions

Related Reading