---
title: "AI Avatar Disclosure: Meta Ads, TikTok Ads and YouTube Uploads"
description: "TikTok requires advertisers to disclose AI content, YouTube requires creators to disclose realistic AI uploads, and Meta labels ads itself. Here is how to submit and recheck an avatar ad."
canonical: https://www.adspecialist.de/en/blog/ai-avatar-ad-disclosure-meta-tiktok-youtube/
last-updated: 2026-11-17T08:01:35.695Z
---

# AI Avatar Disclosure: Meta Ads, TikTok Ads and YouTube Uploads

> TikTok requires advertisers to disclose AI content, YouTube requires creators to disclose realistic AI uploads, and Meta labels ads itself. Here is how to submit and recheck an avatar ad.

Kanonische URL: https://www.adspecialist.de/en/blog/ai-avatar-ad-disclosure-meta-tiktok-youtube/

For AI avatar ad disclosure, Meta, TikTok, YouTube and EU law place duties on different parties. TikTok's ad policy requires the advertiser to disclose significantly edited or AI-generated content, or the ad is rejected or restricted. YouTube requires creators to disclose realistic AI content when they upload it. Meta describes its own _AI info_ labels and says it will begin detecting third-party AI signals. That commercial-ad update does not set a manual labeling duty for advertisers, and Meta says the experience may vary by region because of legal requirements. Organic creator posts have a separate self-disclosure rule. In the EU, the AI Act adds a separate disclosure duty for deepfakes, and German law adds the advertising label on top.

I am Moritz Lambrecht, co-founder of Ad Specialist. The UGC we produce for clients is shot by real creators. This workflow uses platform documentation checked in November 2026. It is not legal advice.

## Who must act on each platform?

On TikTok ads, the advertiser labels the content. On YouTube, the uploading creator declares AI use. On Meta ads, Meta applies the labels described in its policy.

| Platform | Who must act | What triggers it | What happens |
| --- | --- | --- | --- |
| TikTok ads | The advertiser | Significantly edited media or AI-generated content in an ad | AIGC label or own clear disclaimer, caption, watermark or sticker; undisclosed AI content is rejected or restricted |
| YouTube uploads | The creator who uploads | Realistic content that is AI-generated or meaningfully altered | AI use set to Yes in YouTube Studio; repeated non-disclosure may lead to a manual label, content removal or suspension from the YouTube Partner Program |
| Meta ads | Meta | Ads made or significantly edited with Meta's own AI tools; third-party detection announced in the June update | AI info labels; the June update also includes them in About this ad |

Check political and social-issue advertising separately.

## What does Meta do with avatar ads?

Meta's update of 1 June 2026 says the new About this ad area in the three-dot menu will also include AI info labels. It says Meta will begin detecting third-party AI through industry-standard signals and notes that the experience may vary by region because of legal requirements. It does not withdraw the earlier Sponsored-label rule.

Meta's post from 3 February 2025 describes labels for ads made or significantly edited with Meta's own generative AI tools. When those tools add an AI-generated photorealistic human, the label appears next to the _Sponsored_ label, not behind the three-dot menu. Minor edits without a photorealistic human get no label.

Check your placement's labels. Do not strip metadata to evade detection. Organic Facebook and Instagram posts have a different rule: Meta requires its disclosure tool for photorealistic AI video or realistic-sounding AI audio, and says it may penalize failures to disclose. A creator posting an avatar video must check that rule before the post becomes an ad. Where the EU deepfake test applies, the deployer's disclosure duty remains separate.

## What does TikTok require from advertisers?

TikTok's advertising policy on misleading and false content, last updated in November 2026, allows significantly edited media and AI-generated content only if you apply the AIGC label or add your own clear disclaimer, caption, watermark or sticker. It covers fully AI-generated images, video or audio, content that shows the primary subject doing something they did not do, and content that makes the subject say something they did not say through voice cloning. Minor edits, such as lighting or color changes, are outside the rule. When in doubt, TikTok says to label.

If TikTok finds undisclosed AI content, the ad is rejected or restricted. The same policy prohibits content that uses or promotes the misuse of a public figure's likeness without permission, particularly in misleading contexts, and says TikTok removes private persons' likeness used without permission.

Spark Ads run an existing organic post. Before authorizing an avatar post for Spark Ads, check the organic post for the disclosures that apply to its content and commercial context. Our [Spark Ads glossary entry](https://www.adspecialist.de/en/glossary/spark-ads/) explains the format.

## What does YouTube require for creator uploads?

YouTube's help page _Disclosing use of GenAI content_ gives four disclosure triggers: making a real person appear to say or do something they did not do; altering footage of a real event or place; generating a realistic scene that did not occur; and creating music that is the main focus of the video. The creator sets _AI use_ to Yes under Attributes in YouTube Studio during upload. YouTube then adds a label.

Non-realistic content and minor edits need no disclosure. YouTube lists, among others, special-effect filters such as background blur, AI help with scripts or thumbnails, and cloning one's own voice for voice-overs. YouTube may label content automatically, for example content made with its own AI tools or content with C2PA metadata. Creators who repeatedly fail to disclose risk a manual label, removal of content or suspension from the YouTube Partner Program.

Put the creator's _AI use_ duty into the brief and contract. For ads uploaded through Google Ads, check the format's ad policies separately; this help page covers creator uploads.

## What does the EU AI Act add?

Article 50 of the AI Act applies from 2 August 2026. Two roles matter here:

-   **Providers** of generative AI systems must mark outputs in a machine-readable way. For systems placed on the market before 2 August 2026, the Commission's FAQ says this marking duty applies from 2 December 2026.
-   **Deployers** who use AI in a professional activity must disclose deepfakes. According to the FAQ, a business stays the deployer even when contractors operate the tool on its behalf and under its responsibility and control. The 2 December date is not a delay for this duty.

The FAQ defines a deepfake by three cumulative criteria: high resemblance to the simulated subject, a subject that exists, can plausibly exist or could have plausibly existed in reality, and a false appearance of being authentic. A photorealistic avatar presented as a real customer can meet all three. The disclosure must be clear, distinguishable and perceivable at first exposure, for example through a visible or audible label. A deployer cannot rely only on the provider's machine-readable mark. Content generated before 2 August 2026 does not need a label retroactively.

The Commission published guidelines on these obligations on 20 July 2026. A voluntary code of practice from 10 June 2026 sets out marking and labeling measures, and the EU offers icons that deployers may use. Our article on [when AI-generated ad videos need EU disclosure](https://www.adspecialist.de/en/blog/when-ai-generated-ad-videos-need-eu-disclosure/) covers how to classify the asset for the EU layer. The workflow below covers the platform layer.

## How do you submit and recheck an avatar ad?

Run these steps for every avatar ad and every new variant. A new hook or a new language is a new asset.

1.  **Log the asset.** Tool, model, which parts are synthetic (face, voice, scene, product shots) and the file version. Keep the original export with its metadata.
2.  **Classify it.** Is it realistic? Could a viewer take the person or scene as real? Is it a deepfake under the EU test?
3.  **Submit on TikTok.** Turn on the AIGC label or add your own clear disclaimer in the creative before you submit.
4.  **Submit on Meta.** Upload the original export. Where the EU test applies, put a visible disclosure in the creative itself, without relying on Meta's detection or label placement to satisfy the EU duty.
5.  **Brief YouTube creators.** If a creator uploads the video, the creator sets _AI use_ to Yes. Ask for a screenshot of the setting.
6.  **Recheck after approval.** On TikTok, check the review status for a rejection or restriction. On Meta, open _About this ad_ on the live ad. On YouTube, check whether a label appears and whether YouTube Studio shows a notice.
7.  **Recheck on reuse.** When the video moves to Spark Ads or another format, check the source post again.

## Why is an AI label not an ad label?

An AI label tells viewers how the video was made. It does not tell them that a business paid for the message.

The advertising duty comes from section 5a(4) UWG (the binding German text). Where the commercial intent behind a practice is not identified and is not directly apparent from the context, the practice is unfair if that gap is suited to causing a consumer or other market participant to take a transactional decision they would not otherwise take. An activity for another business has no commercial purpose for the benefit of that business if the creator neither obtained nor was promised payment or similar consideration for that activity from the business. The statute presumes that such consideration was obtained or promised, unless the person substantiates that no consideration was received.

If your brand is clearly the sender of a Sponsored ad, the commercial context may already be apparent. A creator posting the video on their own channel presents a different context: make the commercial purpose clear where viewers cannot directly recognize it. Check this separately from the AI label. Never present a synthetic speaker as a real customer who used the product. Have the format reviewed by a lawyer. This is not legal advice.

## Where do real creators fit?

Avatars help when you need many versions of one script. They are weak where the promise is a real person's experience with the product. For creator-shot footage, we book creators to film the product. We test each variant against cost per new customer. Our [UGC page](https://www.adspecialist.de/en/meta-ads/user-generated-content/) shows how that works.

## Frequently asked questions (FAQ)

### What should I save when I check a Meta ad label?

Save the ad ID, the creative version, the check date and a screenshot of the label and About this ad information. Record the account region. Keep this evidence with the approved file so a later edit does not inherit an earlier check.

### How should I hand an AI video to the YouTube uploader?

Give the uploader the final file version and a written note on what was generated or changed. Ask for a record of the AI use setting before publication. Recheck the note if the file changes.

### What should the brief say when a contractor operates the AI tool?

Name who operates the system and who directs that work. Record who checks the legal role and disclosure duties. Do not assign responsibility from the invoice alone; document the actual working arrangement for the review.

### What if the label differs between accounts or regions?

Record the account, region and placement used for each check. Give the reviewer the final file and that record. Resolve a missing disclosure before launch rather than assuming that a label seen in another account will appear.

### How do I record approval for both AI disclosure and advertising disclosure?

Use two approval fields for the same creative version. Name the reviewer, date and result for each field. Leave the asset on hold if either check is unresolved. Reopen both checks when the content or placement changes.

## Sources

-   [Expanding GenAI Transparency for Meta's Ads Products](https://about.fb.com/news/2025/02/gen-ai-transparency-metas-ads-products/), Meta; published 2025-02-03; updated 2026-06-01; accessed 2026-11-15
-   [Misleading and false content (TikTok Advertising Policies)](https://ads.tiktok.com/resources/help/article/tiktok-ads-policy-misleading-and-false-content?lang=en), TikTok for Business; updated 2026-11; accessed 2026-11-15
-   [About Spark Ads](https://ads.tiktok.com/resources/help/article/spark-ads?lang=en), TikTok for Business; updated 2026-06; accessed 2026-11-15
-   [Disclosing use of GenAI content](https://support.google.com/youtube/answer/14328491?hl=en), YouTube Help; accessed 2026-11-15
-   [Transparency obligations under Article 50 of the AI Act](https://digital-strategy.ec.europa.eu/en/faqs/transparency-obligations-under-article-50-ai-act), European Commission; updated 2026-07-24; accessed 2026-11-15
-   [Guidelines on transparency obligations for providers and deployers of AI systems](https://digital-strategy.ec.europa.eu/en/library/guidelines-transparency-obligations-providers-and-deployers-ai-systems), European Commission; published 2026-07-20; accessed 2026-11-15
-   [Code of Practice on Transparency of AI-generated Content](https://digital-strategy.ec.europa.eu/en/policies/code-practice-ai-generated-content), European Commission; published 2026-06-10; accessed 2026-11-15
-   [Spark Ads: Scale TikTok Creator Content](https://www.adspecialist.de/en/glossary/spark-ads/), Ad Specialist; accessed 2026-11-15
-   [When AI-Generated Ad Videos Need EU Disclosure](https://www.adspecialist.de/en/blog/when-ai-generated-ad-videos-need-eu-disclosure/), Ad Specialist; published 2026-10-05; updated 2026-10-05; accessed 2026-11-15
-   [UGC Agency for Meta Ads](https://www.adspecialist.de/en/meta-ads/user-generated-content/), Ad Specialist; accessed 2026-11-15
-   [Act against Unfair Competition (UWG), English translation (non-binding)](https://www.gesetze-im-internet.de/englisch_uwg/englisch_uwg.html), Federal Ministry of Justice and Consumer Protection and Federal Office of Justice (gesetze-im-internet.de); accessed 2026-11-15
-   [Labeling AI-Generated Images on Facebook, Instagram and Threads (updated April 1, 2025)](https://about.fb.com/news/2024/02/labeling-ai-generated-images-on-facebook-instagram-and-threads/), Meta; published 2024-02-06; updated 2025-04-01; accessed 2026-11-15
-   [Act against Unfair Competition (UWG), section 5a (German text)](https://www.gesetze-im-internet.de/uwg_2004/__5a.html), Federal Ministry of Justice and Consumer Protection and Federal Office of Justice (gesetze-im-internet.de); accessed 2026-11-15
