---
title: "Disclosure, AI labels and consent: marketing with AI…"
description: "Three kinds of disclosure marketers deal with When AI is involved in marketing, \"disclosure\" can mean three different things. Mixing them up causes most…"
url: https://optimizeall.com/learn/ai-fundamentals-for-marketers/disclosure-and-ai-labeling
updated: 2026-10-05
---

AI Fundamentals for Marketers & Creators · Privacy, confidentiality and disclosure · lesson 9 of 16 · 13 min

# Disclosure, AI labels and consent: marketing with AI honestly

## Three kinds of disclosure marketers deal with

When AI is involved in marketing, "disclosure" can mean three different things. Mixing them up causes most of the confusion.

| Type | The question | Examples |
|---|---|---|
| **Advertising disclosure** | Is this content paid, sponsored or incentivized? | #ad, "Paid partnership" labels, affiliate disclosures; applies whether or not AI was used |
| **AI-content labeling** | Is this realistic image, video, audio or text made or materially altered by AI? | Platform AI labels, content credentials, EU AI Act transparency duties for deepfakes |
| **Process disclosure** | Did AI help produce this deliverable? | Client contracts requiring it, publisher policies, academic rules |

## Advertising rules apply to AI content too

Regulators judge the **overall impression** an ad creates, however it was produced:

- In the **US**, the FTC's Endorsement Guides require clear disclosure of material connections, and the FTC's rule on fake reviews and testimonials (in force since 2024) prohibits fake reviews and testimonials, including AI-generated ones presented as real customers.
- In the **UK**, the ASA/CAP Code requires ads to be obviously identifiable and not misleading; the ASA has ruled against beauty ads whose filters or edits exaggerated a product's effect, and the same principle applies to AI edits.
- In **Pakistan, the UAE and Saudi Arabia**, advertising and influencer rules (for example the UAE's media and advertising regulations and Saudi Arabia's requirements for licensed advertisers) apply to AI-assisted content the same way; check the current local regulator guidance.

Practical consequences:

- **No AI "customers."** Never present AI-generated people, reviews or testimonials as real customers.
- **No exaggerated results.** An AI-enhanced before/after image that overstates what a product does is misleading, label or no label.
- **A disclosure doesn't cure a false claim.** "Made with AI" next to an untrue performance claim is still an untrue claim.

## AI-content labels on platforms

Major platforms ask creators to label **realistic** AI-generated or significantly altered content (for example, a realistic person saying something they didn't say, or a realistic event that didn't happen), and some apply labels automatically when they detect industry content credentials. Rules and label names change, so check each platform's current help pages (Meta, TikTok, YouTube, LinkedIn). Generally:

- **Label:** realistic synthetic people, voices or scenes; realistic edits that change what happened.
- **Usually no label needed:** AI help with captions, spelling, color correction, obviously stylized or animated content, background clean-up that doesn't mislead. (Check the platform's specifics.)

## The EU AI Act transparency rules

From **2 August 2026**, Article 50 of the EU AI Act introduced transparency duties. For marketers deploying AI content in the EU, the most relevant are: **deepfakes** (realistic AI-generated or manipulated image, audio or video of people, places or events) must be disclosed as such, and AI-generated **text published to inform the public on matters of public interest** must be disclosed unless it has had human review and editorial responsibility. Providers of generative AI tools also have duties to mark outputs in machine-readable ways. If you market into the EU, follow the Commission's guidance and code of practice as they develop; the Responsible AI course covers this in depth.

## Likeness, voice and consent

Creating a synthetic voice or likeness of a real person (a founder, an influencer, a celebrity) needs that person's **explicit, written consent** and a clear agreement on use. Never imitate real people without it, and check local personality-rights and data-protection rules.

## Worked example

A skincare brand in London uses an AI-generated model in a launch ad. The team (1) labels the image in line with the platform's AI label, (2) uses a real, verified before/after photo for any result claim instead of an AI edit, (3) keeps the influencer partnership labeled as "Paid partnership" with #ad, and (4) notes AI assistance in the campaign report because the retailer's contract requires it. The ad runs without complaints, and the brand has a record of every decision.

## Hands-on: the disclosure decision checklist

```text
For each asset, answer:
1. Is it paid, sponsored, gifted or affiliate?           -> use the platform's paid label + #ad (or local equivalent)
2. Does it show a realistic person, voice or event made or materially altered by AI?  -> apply the platform's AI label; for EU audiences, disclose as AI-generated/manipulated
3. Does it show results, reviews or testimonials?        -> real, verified and typical only; never AI "customers"
4. Does it use a real person's likeness or voice?        -> written consent on file
5. Does the client contract or publisher require AI-assistance disclosure?  -> add the agreed wording
6. Is any regulated claim involved (health, finance, legal)?  -> approved wording and sign-off
```

Paste this into your brand project so the assistant appends it to every campaign plan:

```text
For every asset you draft, add a "Disclosure" line stating: paid label needed (yes/no),
AI label needed (yes/no and why), and any claim that needs approval.
```

**Before:** a campaign plan with an AI-generated "happy customer" video and a caption saying "Real results in 7 days!"

**After:** the "customer" is relabeled as an AI-generated spokesperson with an AI label, the results claim is removed pending real evidence, and each asset carries a Disclosure line.

## Pitfalls

- Treating an AI label as permission to exaggerate.
- Forgetting #ad because "the AI wrote the caption".
- Using a celebrity-like synthetic voice "as a joke".
- Assuming last year's platform rules still apply.

## How to measure success

Zero ad rejections or complaints related to disclosure, a Disclosure line on every asset, and consent records for every synthetic likeness.

## Video lecture: Disclosure, AI labels and consent: marketing with AI honestly

Lecture coming soon · 11 chapters · about 9 minutes. Read the full transcript below.

1. Disclosure and AI labels
2. Why it matters
3. Three kinds of disclosure
4. Advertising rules still apply
5. AI labels and the EU AI Act
6. Likeness and voice
7. Example 1: the sponsored Reel
8. Example 2: the AI model campaign
9. Watch me do it
10. Common mistakes
11. Recap and try this now

## Lecture transcript

### Disclosure and AI labels

A brand posts a beautiful ad. A glowing model, a perfect before and after, a caption about real results in seven days. Every element was made or edited with AI. Is that okay? Does it need a label? Which one? And would a label even make it okay? Marketers are asking these questions every day, and the answers are clearer than you might think, once you separate three different kinds of disclosure. In this lecture you'll learn those three kinds, the rules that apply in the US, the UK, the Gulf and the EU, and a six question checklist. You'll see two examples and watch me run the checklist on a real campaign plan.

### Why it matters

Why does this matter? Three audiences are watching. Regulators, who enforce advertising and consumer protection law regardless of how content was made. Platforms, which reject ads, add labels, or reduce reach for content that breaks their rules. And your audience, who increasingly notice synthetic content and feel deceived when it's hidden. The cost of getting it wrong ranges from a rejected ad to a regulator's ruling to a public backlash. The cost of getting it right is a few seconds per asset. And here's the upside: brands that are transparent about AI use tend to keep trust even as AI content becomes everywhere.

### Three kinds of disclosure

Here's the key idea. There are three separate questions. First: is this content paid, sponsored, gifted or affiliate? That's advertising disclosure, like hashtag ad and paid partnership labels, and it applies whether or not AI was involved. Second: does this content show a realistic person, voice or event that AI made or significantly altered? That's AI content labeling, which platforms ask for and which the EU AI Act requires for deepfakes. Third: did AI help produce this deliverable? That's process disclosure, which some client contracts, publishers and institutions require. Most confusion comes from mixing these up. A caption written by AI doesn't need an AI label. But a sponsored post always needs its ad label.

### Advertising rules still apply

Advertising law judges the overall impression an ad creates, however it was produced. In the US, the FTC requires clear disclosure of material connections, and its rule on fake reviews and testimonials prohibits fake reviews, including AI generated ones presented as real customers. In the UK, the ASA requires ads to be identifiable and not misleading, and has ruled against beauty ads using filters that exaggerated a product's effect, a principle that applies just as much to AI edits. In Pakistan, the UAE and Saudi Arabia, local advertising and influencer rules apply to AI assisted content the same way. So three rules of thumb. No AI customers. No exaggerated AI results. And a disclosure never cures a false claim. Made with AI next to an untrue claim is still an untrue claim.

### AI labels and the EU AI Act

Now AI content labels. Major platforms ask creators to label realistic AI generated or significantly altered content, like a realistic person saying something they never said, or a realistic event that never happened, and some apply labels automatically when they detect content credentials. Help with captions, spelling, color correction, or obviously stylized animation usually doesn't need a label, but check each platform's current rules, because they change. And if you market into the European Union, Article fifty of the EU AI Act has applied since the second of August twenty twenty six. Deepfakes must be disclosed as AI generated or manipulated, and AI generated text published to inform the public on matters of public interest must be disclosed unless a human has reviewed it and taken editorial responsibility.

### Likeness and voice

One more rule deserves its own moment: likeness and voice. Creating a synthetic voice or a digital likeness of a real person, whether that's your founder, an influencer you work with, or a celebrity, needs that person's explicit, written consent and a clear agreement about how it will be used, for how long, and where. Never imitate a real person without it, not even as a joke, and not even if the tool lets you. Beyond platform rules, many places have personality rights and data protection laws that apply to someone's face and voice. Consent isn't a formality. It's the whole foundation.

### Example 1: the sponsored Reel

A simple example. A fitness creator in Karachi makes a sponsored Reel for a protein brand. She used AI to help write the script. Which disclosure does she need? The advertising one: the platform's paid partnership label and hashtag ad, or the local equivalent, because it's sponsored. Does she need an AI label? Not for an AI assisted script. She filmed herself, in real life, saying words she chose to say. And the claims in the script, like how much protein is in a serving, must match the product label, because the brand and the creator are both responsible for them. AI or not, the advertising rules are the same.

### Example 2: the AI model campaign

Now a realistic campaign, with illustrative details. A skincare brand in London launches a serum with an AI generated model in its main ad. The team makes four decisions. The image carries the platform's AI label, because it shows a realistic person who doesn't exist. Any results claim uses a real, verified before and after photo, never an AI edit, because an enhanced result would be misleading, label or not. The influencer posts are labeled as paid partnerships with hashtag ad. And the campaign report notes AI assistance, because the retailer's contract requires it. The ad runs without a single complaint, and if anyone asks, the brand has a record of every decision.

### Watch me do it

Let me run the checklist on a real style plan. I'm in a brand project, and I've pasted a five asset campaign plan for a hair care launch. I add the instruction: for every asset, add a disclosure line stating paid label needed, AI label needed and why, and any claim needing approval. Here's what comes back. Asset one, influencer post: paid label yes, AI label no. Asset two, an AI generated happy customer video with the caption real results in seven days: flagged. The AI says this presents a synthetic person as a customer and makes an unverified results claim. Exactly. I change it to a clearly identified AI spokesperson, remove the results claim until we have real evidence, and apply the AI label. The other three assets are fine. Five minutes, and a problem caught before it cost us anything.

### Common mistakes

The common mistakes. Treating an AI label as permission to exaggerate. It isn't. Forgetting hashtag ad because the AI wrote the caption and nobody added it. Using a celebrity like synthetic voice as a joke, which can breach platform rules and someone's rights. And assuming last year's platform rules still apply. Labels, thresholds and tools change regularly, so check each platform's help pages when you plan a campaign, and look at your local regulator's guidance too.

### Recap and try this now

Let's recap. Ask three separate questions: is it paid, is it realistic synthetic media, and does a contract require process disclosure? Advertising rules always apply, so no AI customers, no exaggerated AI results, and remember that a label never cures a false claim. Label realistic synthetic people, voices and events according to platform rules and, in the EU, the AI Act's transparency duties. And get written consent before creating anyone's synthetic voice or likeness. Here's your try this now. Paste the disclosure instruction from the lesson text into your brand project, then run your next campaign plan through it. For the full legal picture, the Responsible AI, Disclosure and Compliance course goes deeper.

## Key takeaways

- Separate three things: advertising disclosure (#ad, paid labels), AI-content labels, and process disclosure required by contracts.
- Advertising law judges the overall impression: no AI 'customers', fake reviews or exaggerated AI results, and a label doesn't cure a false claim.
- Label realistic AI-generated or altered people, voices and events per platform rules; EU Article 50 transparency duties apply from 2 August 2026.
- Get written consent before creating any synthetic voice or likeness of a real person.

## Try it

Add the disclosure instruction to your brand project and run your next campaign plan through the six-question checklist.

- [Previous: Safe AI workflows for client and customer work](https://optimizeall.com/learn/ai-fundamentals-for-marketers/safe-workflows-client-work)
- [Next: Market and audience research with AI](https://optimizeall.com/learn/ai-fundamentals-for-marketers/market-and-audience-research)
- [All lessons of AI Fundamentals for Marketers & Creators](https://optimizeall.com/learn/ai-fundamentals-for-marketers)