How to Label AI-Generated Ads to Comply with New Disclosure Rules in 2027

Learning how to label AI-generated ads to comply with new disclosure rules in 2027 is no longer optional homework for legal departments — it is a production requirement. By January 1, 2027, California’s synthetic-performer disclosure law (SB 1050) and its platform provenance duties (AB 2713) take effect, joining New York’s already-active synthetic performer law, the EU AI Act’s transparency obligations, and California’s rewritten AI Transparency Act. If your 2027 media plan includes any AI-generated people, voices, or imagery in ads, this guide gives you the full rulebook — what each law requires, the exact disclosure wording that satisfies them, and a practical labeling workflow your team can implement before the deadlines hit.

Note: Fast-moving legal area — this reflects statutes and guidance available as of October 2026, for marketing-operations planning, not legal advice. Have counsel review your compliance program.

The 2027 Rulebook at a Glance

Law	Signed	Effective	What it requires
California SB 1050 (synthetic performers in ads)	Sept 16, 2026	Jan 1, 2027	Clear, conspicuous disclosure when an ad prominently features a synthetic performer
California AB 2713 (platform provenance duties)	Sept 30, 2026	Jan 1, 2027	Large platforms must surface provenance data/digital signatures; may not knowingly strip them
California SB 1000 (AI Transparency Act rewrite)	Sept 30, 2026	Immediately (urgency statute)	AI providers must embed hidden provenance data in generated files and offer a free verification tool
New York GBL § 396-b (synthetic performers in ads)	2025 session	June 9, 2026 (in force)	Disclosure when advertiser knows a synthetic performer is used
EU AI Act, Article 50	2024	Aug 2, 2026 (in force)	Deployers must disclose AI-generated/manipulated deepfake content; providers must machine-mark outputs
FTC Act / Endorsement Guides	Ongoing	In force	No deceptive ads; testimonials must be honest; material connections disclosed

California SB 1050: The Law That Changes Ad Production

SB 1050 is the centerpiece for advertisers. Signed by Governor Newsom on September 16, 2026 and effective January 1, 2027, it makes California the second state after New York to require disclosure of synthetic performers in advertising.

What counts as a synthetic performer? A digital figure, voice, or representation created in whole or in part with generative AI that creates the realistic impression of a human performer who is not recognizable as any identifiable real person — wholly fictional digital humans. A licensed digital replica of a real celebrity is not a “synthetic performer” under this definition (California regulates replicas separately).

When does the duty trigger? Only when the synthetic performer appears “prominently” — the statute defines that as three situations:

1. The figure appears in the foreground demonstrating or illustrating the product or service;
2. It provides or voices the on- or off-camera narration or commercial message; or
3. It illustrates or reacts to that message.

A synthetic person visible only as a blurred background extra likely isn’t “prominent.” Your AI spokesperson reading the script absolutely is.

The required wording. The disclosure must be “clear and conspicuous” and use wording substantially similar to one of two statutory examples:

• “This performance features a synthetic performer”, or
• “No human performer is depicted.”

Using California’s wording on every version of a national campaign is the simplest approach — it satisfies New York’s parallel requirement too.

Exemptions. Ads for expressive works (films, TV, streaming, games) are exempt where the synthetic performer’s use matches the underlying work. Audio-only ads are covered — California goes further than New York here.

Enforcement — the part that gets attention. Violations are false-advertising violations (B&P § 17500), enforceable under the Unfair Competition Law (§ 17200): by the Attorney General, district attorneys, certain city attorneys — and by private plaintiffs. That private-enforcement path makes California exposure materially higher than a slap on the wrist.

New York’s Law: Already Live, Different in the Details

New York’s GBL § 396-b has been in force since June 9, 2026 — so national campaigns are already covered there. Two differences from California matter operationally: New York’s duty bites only where the advertiser had actual knowledge a synthetic performer was used (California has no such qualifier), and New York exempts audio-only ads while California covers them. Neither law requires a materiality showing — the label is due regardless, unlike FTC deception claims.

SB 1000 and AB 2713: The Provenance Layer Underneath the Label

Visible labels are only half the 2027 story. California also rebuilt the machine-readable layer:

SB 1000 (signed September 30, 2026; effective immediately as an urgency statute) rewrites the California AI Transparency Act (originally SB 942). What changed for advertisers to know:

• AI providers must embed latent (hidden) provenance disclosure in every AI-generated image, video, or audio file, stating whether the AI created or altered the content.
• Providers must offer a free disclosure verification tool anyone can use to check that data.
• The old one-million-monthly-user threshold is gone — any generative AI system publicly accessible in California is now covered.

AB 2713 (signed September 30, 2026; operative January 1, 2027) puts duties on large online platforms: they must let users see whether provenance data or a digital signature is available and inspect it (via UI display, download, or link), and they are barred from knowingly stripping standards-compliant provenance data from content uploaded, distributed, or downloaded on the platform, to the extent technically feasible.

What this means for your workflow: archive original AI-generated files with metadata intact before editing. Re-encoding or stripping metadata destroys the provenance chain — and after January 1, 2027, platforms must not knowingly strip it either.

The EU AI Act: Article 50 Applies to Ads Too

Since August 2, 2026, the EU AI Act’s Article 50 transparency duties apply — including to US companies whose AI-generated content is used in the EU. For advertisers:

• Deployers (the brand publishing the ad) must disclose AI-generated or manipulated deepfake content. The Commission’s July 2026 guidelines interpret “deepfake” broadly — realistic AI-generated people, objects, places, or events that falsely appear authentic, which covers most AI ad creative.
• The disclosure must be clear, distinguishable, given at the latest at first exposure, and accessible. For video: an on-screen notice at the start and after interruptions. For audio: an audible disclaimer at the beginning. For images: a visible label, e.g., top-right, plus metadata.
• The Commission provides free official EU icons (“AI GENERATED” / “AI MODIFIED”) that performed well in user testing — the simplest uniform compliance visual.
• Providers must mark outputs in machine-readable format (deadline December 2, 2026 for systems already on the market). If your ads reach EU audiences, add the EU icon alongside your US wording — one creative, two labels, full coverage.

The Practical Labeling Playbook: From Asset Audit to Launch

Step 1: Audit Every Creative Asset (Q4 2026)

Build a register of every ad asset: for each, record whether it contains a synthetic performer, an AI-generated voice, AI-generated imagery, or AI-manipulated real footage — and require vendors to declare AI use in writing. You cannot label what you haven’t inventoried.

Step 2: Apply the Right Label per Format

• Video ads: on-screen text for the full duration the synthetic performer is prominent (minimum: opening 3+ seconds and end card), e.g., “This performance features a synthetic performer.” Plus the platform’s AI-content toggle at upload and provenance metadata preserved in the file.
• Audio ads: audible disclaimer at the start — California covers audio-only, New York doesn’t, so label for California and you’re covered in both.
• Static/display ads: visible “AI-generated” label on the creative itself (not a footnote or terms page), plus metadata tags.
• UGC-style and influencer ads: real creators filmed on camera are not synthetic performers — SB 1050 doesn’t apply. If the “creator” is AI-generated, label it; FTC paid-relationship disclosure still applies either way.

Step 3: Lock the Pipeline

• Archive untouched originals with provenance metadata before any edit.
• Add the disclosure as a template element (lower-thirds, end cards, audio beds) so no version ships without it.
• For EU placements, add the official EU “AI GENERATED” icon with alt text.

Step 4: Calendar the Remaining Deadlines

New York’s law and SB 1000’s provider duties are already in force. EU provider machine-marking is due December 2, 2026. California SB 1050 + AB 2713 platform duties take effect January 1, 2027 — have labeled creatives live before that date.

4 Labeling Mistakes That Create Liability

1. Burying the label. A disclosure in the footer, terms page, or video description alone isn’t “clear and conspicuous” — it must be visible at first exposure, without interaction.
2. Labeling only some versions. The 9:16 cut, the 6-second bumper, the audio-only version — every variant that prominently features the synthetic performer needs the label.
3. Assuming the platform toggle is enough. YouTube/Meta/TikTok AI labels help but don’t satisfy SB 1050’s statutory wording or the EU’s deployer duties.
4. Thinking a label fixes a misleading ad. A label doesn’t make an otherwise deceptive ad acceptable — fabricated endorsements and exaggerated results remain illegal, labeled or not.

Frequently Asked Questions

What exact wording should I use to label AI-generated ads in 2027?

For US campaigns, use California SB 1050’s statutory phrasing: “This performance features a synthetic performer” or “No human performer is depicted.” It satisfies both California (from January 1, 2027) and New York (already in force). For EU audiences, add the official EU “AI GENERATED” icon.

When does California SB 1050 take effect?

January 1, 2027. It was signed September 16, 2026. New York’s equivalent law has been in force since June 9, 2026, so national campaigns need labels now, not in January.

Does SB 1050 apply to AI voiceovers in ads?

Yes — the definition covers digital figures, voices, or representations, and California explicitly reaches audio-only advertising. An AI voice narrating your commercial is a synthetic performer if it creates the realistic impression of a human performer.

What is the difference between SB 1000 and SB 1050?

SB 1050 (effective Jan 1, 2027) requires visible disclosures on ads featuring synthetic performers. SB 1000 (in force since Sept 30, 2026) requires AI providers to embed hidden provenance data in generated files and offer a verification tool. One is the label viewers see; the other is the metadata machines read.

What does AB 2713 require from platforms starting January 1, 2027?

Large online platforms must let users see and inspect provenance data or digital signatures on content, and must not knowingly strip standards-compliant provenance data from content on their platforms where technically feasible.

Do I need to label AI-generated ad copy or backgrounds?

Under California’s ad laws, the trigger is a prominently featured synthetic performer — routine AI-assisted copy, color correction, or decorative backgrounds don’t trigger SB 1050. But EU AI Act duties are broader for realistic depictions, and FTC deception principles always apply. When in doubt, label.

Need Compliant AI Ad Creatives? This guide is the DIY path — and it works. But if you’d rather have an experienced team produce your AI ad creatives with compliance built in — proper disclosures, provenance-preserving pipelines, and platform-native formats — NAN Marketing Agency does exactly that: AI-powered video ads and social content for brands worldwide. WhatsApp: +92 317 7415702 | Email: nidahina99@gmail.com | Website: https://nidaayesha.com