Responsible AI, Disclosure & ComplianceLabeling AI content and advertising disclosure · Lesson 6 of 11
Advertising and influencer disclosure when AI is involved
Video lecture
Advertising and influencer disclosure when AI is involved
The narrated lecture is in production
Every chapter is scripted and ready. Browse the chapters and read the full transcript now — the video will appear here when it’s published.
Chapters
Transcript of the narration, chapter by chapter.
0:00 Ad disclosure when AI is involved
Here's a caption an AI tool wrote for a sponsored skincare post: clinically proven to erase wrinkles in seven days. It sounds great. It's also, very probably, a problem, and the creator hasn't even decided whether to say it's an ad yet. In this lecture you'll learn the two separate disclosure duties, what the FTC and the UK's ASA actually expect, how the UAE and Saudi Arabia license influencer advertising, the new risks AI creates, and a compliance gate you can put between your AI tool and the publish button.
0:39 Why it matters
Why does this matter? Because when AI meets paid content, you usually have two disclosure duties at once, and mixing them up is one of the most common mistakes in the industry. Get it wrong and you risk regulator action, fines in some markets, platform penalties, and brands dropping you. Get it right and you become the creator or agency brands trust with bigger budgets, because you make their legal team's life easier.
1:11 Two separate duties
Here's the key idea. Think of a packaged snack. The nutrition label tells you what's in it. A separate stamp tells you it's on promotion. One doesn't replace the other. Same here. Commercial disclosure tells people you're being paid, got free product, or earn commission. AI disclosure tells people realistic content is AI-generated or altered, where that's required or where leaving it out would mislead. A sponsored Reel with an AI voiceover may need a clear ad label, the paid partnership tool, and an AI label.
1:48 United States: FTC
Let's go deeper on the US. The FTC's Endorsement Guides, revised in twenty twenty-three, require disclosure of material connections that is clear and conspicuous: hard to miss and easy to understand. In video, that means saying it and showing it in the video, not just the description. The FTC has also said platform paid-partnership tools may not be enough on their own. And since October twenty twenty-four, the FTC's rule on consumer reviews and testimonials prohibits fake reviews and testimonials, including ones from people who don't exist or generated by AI, with civil penalties available. If you advertise an AI product, its capability claims need evidence too.
2:35 United Kingdom: ASA, CAP, CMA
Now the UK. The CAP Code, applied by the ASA, says ads must be obviously identifiable, which in practice means ad or hashtag ad, upfront. Vague tags like sp, collab or gifted aren't enough for paid content. The ASA's position on AI is simple: there's no AI-specific rule, and ads made with AI are judged by the same rules as any other ad. You must disclose AI use where leaving it out would mislead. But here's the crucial point: a disclosure doesn't cure a misleading ad. An AI image that exaggerates a cream's effect is misleading even with an AI-generated note. And the CMA enforces the ban on fake reviews under the twenty twenty-four consumers act, in force since April twenty twenty-five.
3:28 Gulf and Pakistan
In the Gulf, licensing comes first. In the UAE, the Media Council requires an Advertiser Permit for people who promote products or services on social media, under the twenty twenty-three media law; it became mandatory after a grace period that ended in January twenty twenty-six, with some exemptions, like promoting your own business. In Saudi Arabia, the General Authority for Media Regulation, which replaced the older audiovisual media commission, issues the Mawthooq license for people who earn from advertising on social media. In both, ads must be clearly identified. For Pakistan, consumer protection, cybercrime rules and platform policies apply, so clear, honest disclosure is the safe standard. Always check current rules before a campaign.
4:18 Where AI adds risk
Where does AI create new risk? Five places. AI-generated testimonials or reviews presented as real customers: prohibited in the US and UK. Virtual influencers: they must disclose the paid relationship and can't imply they genuinely tried a product. Cloned voices or avatars of real people implying endorsement without consent: false endorsement. Exaggerated visuals, like AI-perfected skin or oversized food portions. And auto-generated variants: two hundred AI versions of an ad can slip in claims nobody approved. Every variant must meet the same standard as the original.
4:55 Example 1: Karachi food creator
First example, simple. A Karachi food creator is paid to promote a new burger. The brand sends AI-generated images of an enormous, perfect burger. She shoots the real product instead, in her kitchen, starts the caption with ad, paid partnership with the brand, and uses the platform's paid partnership tool. One AI-generated background shot in her intro is realistic, so she adds a line: intro background is AI-generated. Two duties, both done, and her comments fill with people saying it looks exactly like what they got.
5:32 Example 2 (illustrative): skincare, UAE + US
Second example, a realistic business scenario with illustrative details. A US skincare brand briefs a Dubai-based creator for a campaign aimed at UAE and US audiences. The agency's AI tool generates forty caption variants. One says clinically proven to erase wrinkles. The brand's approved claim is: helps improve the appearance of fine lines in four weeks, based on a brand consumer study. The compliance gate catches it and blocks it. The agency confirms the creator's UAE Advertiser Permit, every caption opens with ad, the paid partnership tool is on, and a shot of AI-generated laboratory footage is labeled as an illustrative AI visual. Illustratively, legal signed off in one round instead of three, and the campaign launched on schedule.
6:24 Watch me do it: the compliance gate
Watch me do it. I paste the brand's approved claims into the compliance prompt from the lesson: helps improve the appearance of fine lines in four weeks, and fragrance-free. Then I paste six variants with IDs. The model returns a table. Variant two quotes clinically proven to erase wrinkles: BLOCK. Variant four has no commercial disclosure: FIX. Variant five says real customers love it, next to AI faces: fake testimonial risk, BLOCK. The rest pass. Now the important part: I, a human, read the table, check variants two and five myself, fill in the sign-off record, and only then do the passing variants go to scheduling.
7:10 Common mistakes
Common mistakes. Believing an AI label covers commercial disclosure. It doesn't. Letting AI generate ad variants without a claims review. Virtual influencers implying real-life product experience. Relying only on the platform's paid partnership tag with no words. Posting paid content for UAE or Saudi audiences without checking the creator's permit or license. And adding an AI-generated note to an exaggerated image as if that fixes it. It doesn't. Show real results, or don't show results.
7:43 Recap + try this now
Recap. Commercial disclosure and AI disclosure are separate duties; you often need both. The FTC wants clear and conspicuous disclosure and bans fake and AI testimonials. The ASA wants ads obviously identifiable and judges AI ads by the same rules; a label doesn't fix a misleading ad. The UAE and Saudi Arabia license people who advertise on social media. And AI variants need a claims gate plus human sign-off. Try this now: run your last sponsored post or client campaign through the checklist and compliance prompt in the lesson, and fix anything it finds. Next module: EU transparency law and content provenance.
Two separate duties
When AI meets paid content, you usually have two separate disclosure duties:
- Commercial disclosure: telling audiences when content is advertising, sponsored or involves a material connection (payment, free products, affiliate commission, family or business ties).
- AI disclosure: telling audiences when realistic content is AI-generated or altered, where required or where it would otherwise mislead.
One does not replace the other. A sponsored Reel with an AI voiceover may need both "#ad" and the paid-partnership label, and an AI label.
Commercial disclosure principles
Rules differ by country, but the core principles are consistent:
- Clear and prominent: at the start of the caption or visible in the video, not buried among hashtags or behind "more".
- Unambiguous wording: "#ad", "Ad", "Advert", "Paid partnership with [brand]". Vague tags like "#sp", "#collab" or "thanks to [brand]" are often considered insufficient.
- Use platform tools such as paid-partnership labels in addition to, not instead of, clear wording where regulators expect it.
- Every format: stories, livestreams, short video, podcasts and newsletters.
- Truthful claims: endorsements must reflect genuine opinions and experience; claims must be substantiated.
Regional snapshot (principles; check current rules)
- United States: the FTC Endorsement Guides (revised 2023) require clear and conspicuous disclosure of material connections, and the FTC's Trade Regulation Rule on Consumer Reviews and Testimonials (in effect since October 2024) prohibits fake reviews and testimonials, including AI-generated ones, and misrepresented endorsements, with civil penalties available.
- United Kingdom: the CAP Code requires ads to be obviously identifiable, with "#ad" upfront commonly expected; the CMA enforces consumer protection law, including the ban on fake reviews under the Digital Markets, Competition and Consumers Act 2024, in force since April 2025.
- UAE: the UAE Media Council regulates advertising under Federal Decree-Law No. 55 of 2023 on media regulation. People who promote products or services on social media need an Advertiser Permit, which became mandatory after a grace period that ended on 31 January 2026 (exemptions exist, for example for promoting your own business); ads must be clearly identified.
- Saudi Arabia: the General Authority for Media Regulation (GAMR, the successor to the former General Commission for Audiovisual Media) issues the Mawthooq license required for people who earn from advertising on social media, and ads must be clearly disclosed.
- Pakistan: consumer protection, cybercrime and broadcasting rules apply, and platform policies are key; the principle of clear, honest disclosure is the safe standard.
- EU: the Unfair Commercial Practices Directive, the Digital Services Act and national influencer rules require clear commercial disclosure.
Where AI creates new advertising risks
- AI-generated testimonials or reviews: fabricating reviews or "customer" videos with AI avatars is deceptive and specifically prohibited in several markets.
- Synthetic influencers and virtual creators: fully AI personas promoting products should be clear that they are virtual and that the content is paid. They cannot have genuinely "tried" a product, so avoid implying personal experience.
- Cloned voices or avatars of real people implying endorsement without consent: false endorsement.
- Exaggerated visuals: AI-enhanced product images showing results the product cannot deliver, such as flawless skin from a cream or oversized food portions, can be misleading. Some regulators already scrutinize filters on beauty ads.
- Auto-generated ad variations: hundreds of AI variants can introduce unapproved claims. Every variant must meet the same standards.
A compliance checklist for AI-assisted sponsored content
Worked example
A US skincare brand briefs a Dubai-based creator for a campaign across UAE and US audiences. The agency's AI tool generates caption variants, one of which says "clinically proven to erase wrinkles". The checklist catches it: the brand's approved claim is "helps improve the appearance of fine lines in 4 weeks (brand consumer study)". The creator's UAE advertising permit is confirmed, captions open with "#ad", the paid-partnership tool is used, and a shot using AI-generated b-roll of a laboratory is labeled "illustrative AI visual". The creator only posts after actually using the product.
Going deeper: what the FTC and ASA actually expect
FTC (United States)
- Clear and conspicuous means difficult to miss and easily understandable by ordinary consumers. In video, disclose in the video itself (spoken and on screen), not only in the description. In live streams, repeat it.
- Platform tools are not automatically enough. The FTC has said built-in paid-partnership tools may not be sufficient on their own; use clear words as well.
- Consumer Reviews and Testimonials Rule: prohibits writing, selling or buying fake reviews and testimonials, including those attributed to people who do not exist or generated by AI; buying positive or negative reviews conditioned on sentiment; and misrepresenting company-controlled review sites as independent. Civil penalties can apply per violation.
- AI claims themselves: under its "Operation AI Comply" enforcement sweep in 2024 the FTC acted against deceptive AI claims. If you advertise an AI product, claims about what it does need evidence like any other claim.
ASA and CAP (United Kingdom)
- Obviously identifiable: "#ad" or "Ad" upfront; vague labels such as "#sp", "#collab" or "gifted" alone are not enough for paid content.
- No AI-specific rule, same rules apply: CAP's guidance says ads made with AI are judged by the same rules on misleadingness, harm and offense as any other ad. There is no blanket duty to say "made with AI", but you must disclose where leaving it out would mislead.
- Disclosure does not cure a misleading ad: an AI image that exaggerates a cosmetic's effect is misleading even with an "AI-generated" note. Show real results or do not show results.
- Filters and edits: the ASA has repeatedly ruled against beauty ads using filters that exaggerate efficacy; the same logic applies to AI enhancement.
Hands-on: a claims-and-disclosure gate for AI ad variants
When a tool generates dozens of variants, put a gate between generation and publishing. This prompt makes an LLM do the first pass; a human signs off.
You are a compliance pre-checker for ad copy. Approved claims (the ONLY
claims allowed) are listed under APPROVED. Markets: UK, US, UAE.
For each variant under VARIANTS, output a table:
variant_id | unapproved or exaggerated claims (quote them) |
missing commercial disclosure (yes/no) | fake-review or testimonial risk
(yes/no) | AI-disclosure needed (yes/no, why) | verdict (PASS / FIX / BLOCK)
Rules: do not rewrite the ad; do not invent claims; if unsure, mark FIX.
APPROVED:
- "Helps improve the appearance of fine lines in 4 weeks (brand consumer study, n=120)"
- "Fragrance-free"
VARIANTS:
[paste variants with IDs]HUMAN SIGN-OFF RECORD
Campaign: ____ Variants reviewed: __ PASS: __ FIX: __ BLOCK: __
Commercial disclosure wording used: ________________________
Platform paid-partnership tool on: yes / no
Creator license/permit checked (UAE Advertiser Permit / KSA Mawthooq): yes / n.a.
Reviewer: ____ Date: ____Pitfalls
- Believing an AI label covers commercial disclosure.
- Letting AI generate ad variants without claim review.
- Virtual influencers implying real-life product experience.
Key takeaways
- Commercial disclosure and AI disclosure are separate duties, and you often need both.
- Commercial disclosures must be clear, upfront and unambiguous across every format.
- Rules vary by market, including FTC, ASA/CAP and CMA, the UAE Advertiser Permit and the Saudi Mawthooq license from the General Authority for Media Regulation, so check local requirements.
- AI-generated reviews, false endorsements and misleadingly enhanced visuals are key AI advertising risks.
Check your understanding
Quick questions to lock in the lesson. They don’t count towards your certificate.
Put it into practice
Apply the compliance checklist to your last sponsored post or client campaign and fix any gaps you find.
Enrol for free to save your progress
Reading is always free. Enrol to keep your place, take the final assessment and earn a verifiable certificate.