AI advertising · Compliance

Artificial Intelligence Ads Disclosure: Labels, Policies and Compliance by Country

The rule changes at each border, and the platforms add their own labels on top. This is a working map for brands that sell in more than one market.

Ajay Kumar • •9 min read
Illustration of an ad carrying an AI disclosure label

Quick answer: do artificial intelligence ads need a disclosure label?

Disclosure rules for artificial intelligence ads depend on the market. The EU requires disclosure of deepfake image, audio and video. New York requires ads to identify synthetic performers. India requires labelling of AI-generated content on platforms. The UK has no blanket rule. Google and Meta add their own labels, which do not replace legal duties.

The rule in each market, at a glance

No single rule covers AI in advertising. A brand selling in India, the Gulf, the UK, Europe and the US meets a different position in each. This table covers the rules we read at source.

MarketPositionStatus
European UnionDeepfake image, audio or video made or changed with AI must be disclosedApplies since 2 August 2026
New YorkAds must identify AI-generated synthetic performersIn effect since June 2026
CaliforniaDuties on large AI system providers now, and on large platforms laterOperative 2 August 2026
IndiaPlatforms must ensure clear labelling and traceable metadata for AI-generated contentAmended IT Rules, reported July 2026
United KingdomNo blanket duty to disclose. Existing ad rules on misleading content applyRegulator advice, May 2025
United States, federalNo fake reviews or testimonials, however they are madeIn effect since 21 October 2024

This is a summary for marketers, not legal advice. For markets not listed here, such as the UAE and Saudi Arabia, we did not review local law. Platform rules still apply there.

New to the subject? Start with artificial intelligence ads: what they are and how they are made.

Country by country: what each rule asks of an advertiser

European Union

Article 50 of the AI Act puts the duty on whoever uses AI to create or change image, audio or video that counts as a deepfake. They must disclose that it is artificial. The European Commission's guidelines on Article 50 list exceptions, such as standard editing.

In practice: a generated spokesperson needs a label. A colour-corrected product photo does not.

United States

There is no single federal AI label law for ads. Three rules matter most.

India

The government's amendment to the IT Rules deals with synthetically generated information. It requires platforms to ensure clear labelling and traceable metadata for AI-generated content. The duty sits with platforms first, and both Google and Meta now list India among the regions for AI declarations.

United Kingdom

The UK's ad regulator takes a different line. In its advice on disclosing AI in advertising, it says the advertising codes contain no AI-specific rules, and that there is no blanket legal requirement in the UK to disclose AI in ads. The existing rules apply however the content was made.

It suggests two questions. Is the audience likely to be misled if the use of AI is not disclosed? And does the disclosure clarify the ad's message, or contradict it?

South Korea and Vietnam

Google lists both among the regions where AI rules require labels on certain ads. We have not read those laws ourselves, so take local advice before you run AI-made ads there.

How platform labels work on Google and Meta

Platforms do not wait for you to work out each law. They have built their own label systems, and these act in three ways.

Labels you switch on

Google's AI label setting lets you mark an asset as created or edited with AI. Ads using that asset then carry a disclosure in the "How this ad was made" section of My Ad Center, everywhere in the world.

For campaigns targeting the European Union, India, New York, California, South Korea or Vietnam, the ad also shows a visible overlay.

Meta lets advertisers self-disclose in regions with their own AI rules. It currently lists the European region, California, New York, India and Taiwan. The label appears next to the Sponsored label.

Labels the platform adds

Meta labels ads automatically when you use its generative tools to create or significantly edit an image or video, and when it detects standard metadata such as C2PA from outside tools.

Google says it may label assets itself where it is legally required to, when it receives signals from other platforms, or when you use its fully automated creative features. These labels cannot be overwritten.

Labels you design

You can also put your own AI label on an image or video. Google says this does not breach its rules on text overlays and watermarks. It gives two practical tips:

A setting is not legal cover

Google states that using its AI label setting does not guarantee compliance with any regulation. The legal duty stays with the advertiser.

How generation and labels work inside Performance Max is covered in AI ads on Google.

A labelling workflow for brands in several markets

The rules differ, but one routine can serve all of them. Sort every new asset before it reaches a campaign.

  1. Sort. Place each asset in one of three groups: no AI, AI-edited, AI-made.
  2. Flag realism. Mark any AI asset that shows a realistic person, voice or event.
  3. Check truth. Product, claims and any testimonial must be real. If not, stop here and fix the ad.
  4. Label flagged assets in every market. One rule is easier to run than six.
  5. Set the platform declaration for other AI-made assets in regions the platform lists.
  6. Record it. Note the tool, the approver, the markets and the label decision.

What the routine costs

Teams skip this because it sounds slow. Measured, it is small.

Worked example: 60 new assets a month on ₹30 lakh of ad spend

Step 1: Count the work

Example: of 60 new assets, 18 are AI-made or AI-edited, and 7 of those show realistic people or scenes.

Step 2: Time it
Review time per month
Sorting: 60 × 2 minutes = 120 minutes
Detailed check: 18 × 5 minutes = 90 minutes
Total: 210 minutes, or 3.5 hours
Step 3: Price it
Cost against ad spend
3.5 hours × ₹1,200 = ₹4,200
₹4,200 ÷ ₹30,00,000 = 0.14% of spend

Seven assets get a label in every market. The other 11 get a platform declaration where the platform asks for one.

Our rule of thumb

If you are unsure whether a viewer could take an AI scene for a real one, label it. The label costs little. An ad that fools people can cost the account.

The 4-question test behind step 2 is explained in our guide to AI advertising ethics and disclosure.

What a label cannot fix

A disclosure tells people how an ad was made. It does not make a false message true.

The UK regulator puts this clearly. Disclosure alone is very unlikely to repair a fundamentally misleading message. Its example is an AI-generated image showing the effect of a cosmetic product that does not reflect real results. Adding "made with AI" does not rescue that ad.

The same logic runs through the US rule on testimonials. An invented customer is banned whether a person or a model wrote the words.

So keep the order right:

  1. Is the ad true? Product, results and people.
  2. Could the AI content mislead? If yes, change it or disclose it.
  3. Does the market or platform need a label? Apply it.

Whether AI-made ads perform as well as filmed ones is a separate question. What the tests say is in AI ads vs human-made ads.

What most guides get wrong about AI ad disclosure

Many compliance articles copy each other, and several of their claims are out of date or too broad.

A monthly check to stay current

These rules moved several times in 2026 and will move again. A short monthly check keeps you ahead of them.

ScaleWin shows Meta and Google campaigns on one profit view, so you can see what each ad earns before and after a change.

Key takeaways

  • The common target everywhere is realistic AI-made people, voices and scenes. Routine editing is not.
  • EU Article 50 has applied since 2 August 2026, and New York's synthetic performer law since June 2026.
  • The UK regulator says there is no blanket legal duty to disclose AI in ads, but misleading ads break the rules either way.
  • Google shows visible AI labels on ads in six regions when you mark an asset as AI-made. Some platform labels cannot be removed.
  • A label never repairs a misleading claim. Fix the ad first.

Artificial intelligence ads: frequently asked questions

Do AI-generated ads have to be labelled?

Only in some cases. The rules we read target realistic AI-made people, voices and scenes. The EU requires disclosure of deepfake image, audio and video, and New York requires ads to identify synthetic performers. Routine editing, such as cropping or colour correction, is not the target.

Is there a law in the UK requiring AI disclosure in ads?

No blanket one. The UK advertising regulator says its codes have no AI-specific rules and that there is no general legal requirement to disclose AI in ads. Existing rules still apply, so an ad that misleads people breaks them whether AI was used or not.

What does India require for AI content in ads?

India amended its IT Rules to cover synthetically generated information. Platforms must ensure clear labelling and traceable metadata for AI-generated content. Google and Meta both list India among the regions where advertisers can or should declare AI-made assets.

How do I add an AI label in Google Ads?

Open the asset in the Asset Library and choose to label it as created or edited with AI. Google then shows a disclosure in My Ad Center worldwide, and a visible overlay on the ad in the European Union, India, New York, California, South Korea and Vietnam.

Does a platform AI label make my ad legally compliant?

No. Google says plainly that using its AI label setting does not guarantee compliance with specific regulations, and advises advertisers to seek legal guidance. The label is a tool. The duty to follow each market's law stays with the advertiser.

Can a platform label my ad as AI without asking me?

Yes. Meta labels ads made or heavily edited with its generative tools and those carrying standard metadata such as C2PA. Google may label assets where it is legally required to or when its automated features created them. These labels cannot be overwritten.

Does an AI label fix a misleading ad?

No. The UK regulator says disclosure alone is very unlikely to repair a fundamentally misleading message, such as an AI image showing product results nobody achieved. Make the ad accurate first, then decide whether it needs a label.

Sources

Ajay Kumar
Founder, ScaleWin · 12+ years in performance marketing

Ajay is the founder of ScaleWin. He works hands-on with Meta Ads, Google Ads and SEO for Indian D2C and service brands, and built ScaleWin so that the daily work of a performance marketer - reading numbers, pausing losers, scaling winners - happens automatically.

Drafted with AI assistance. How these articles are written

Let ScaleWin read your ad data every day

ScaleWin checks your Meta and Google Ads daily, judges every ad on profit and tells you exactly what to change.

Start free, no card

Get marketing tips that grow your business

One practical email a week on ads, SEO and growth. No spam, no fluff. Unsubscribe anytime.