UAE advertiser permit rules: what influencers and brands in Dubai must check
Who needs a UAE advertiser permit, how the visitor permit differs, and the advert rules brands in Dubai must follow when they pay creators.
Read the articleDigital Marketing Services · AI Ads
A plain explanation of where AI genuinely speeds up paid ad creative in Dubai, where it still falls short, what still needs a person, and what the disclosure rules on Meta, Google and TikTok actually require.
AI ads is the least precise phrase in Dubai advertising right now, because it is used for at least four different things: an image generated from a text prompt, a video actor who does not exist, ad copy drafted by a language model, and a bidding algorithm quietly rebalancing spend between platforms. Some of that is genuinely useful and has been running inside ad platforms for years without anyone calling it AI. Some of it is closer to hype, sold with confidence and delivered with a noticeably synthetic look. This page sets out what AI actually does well in a Dubai paid ad campaign today, what it still does badly, what a person still has to do regardless of which tools are involved, and the disclosure and consent rules that apply the moment a generated asset goes live as a paid advertisement rather than staying in a drafts folder.
If what you need is the wider campaign around the creative, rather than the creative itself, the platform specific pages cover that in full: PPC advertising for search and shopping campaigns, Google Ads management, and social media advertising for the account setup, targeting and measurement side of Meta, TikTok and the other platforms. This page is specifically about the ad creative itself, where AI sits inside making it, and the honesty and disclosure questions that come with that.
Who actually asks for this
The request rarely starts as “we want AI”. It starts as a specific, practical problem that AI tools happen to help with.
Running several ad accounts and needing more headline, image and video variants to test each week than a traditional design and shoot schedule can realistically produce, without lowering the quality bar on what actually gets published.
Needing a first set of credible ad creative before committing to a full photography or video shoot, to validate a product or an offer before spending on production for something that might not work.
Needing the same campaign adapted into Arabic and English, and into several aspect ratios for different platforms, faster than manually resizing and retranslating every asset by hand.
Two different questions
People land here from two quite different starting points, and the useful part of this page is different for each.
You have seen the term and want a straight answer on what it covers, what it is good at, and what to be sceptical of, before deciding whether it belongs in your next campaign at all.
You already run paid ads and want faster, lower effort first drafts, more variants to test, or help with a specific format, and you want to know our actual process, what you need to supply, and the rules that apply once a generated asset is live.
What actually works
Set against the hype, the honest list of what current AI tools are reliably good at is shorter than the marketing around them suggests, but it is real. Language models are good at producing a large number of headline and body copy variants quickly, giving a human editor more raw material to choose from and refine, rather than starting from a blank page each time. Image generation tools are good at fast concept visuals: exploring a direction, a mood or a layout before committing design time to it, and at background and scene generation where no real product or person needs to appear accurately. This is the sensible starting point for most AI ads Dubai briefs that begin with a tight budget and a short timeline.
Resizing and reformatting one approved creative into the many aspect ratios a modern campaign needs, square, vertical, landscape, story format, is a task AI assisted tools now do faster than manually rebuilding each size, provided a person checks that nothing important has been cropped or distorted in the process. Translation and localisation drafts, from English into Arabic or the reverse, are a genuinely useful starting point too, on the strict condition that a native reader reviews the result before it goes anywhere near a live ad account. Treated this way, AI ads for a Dubai audience stay a genuine time saver rather than a source of embarrassing mistakes.
Where it still falls short
This is the part most agencies selling AI ads skip. We would rather say it plainly than have a client find out after a campaign underperforms.
Generated images of a specific real product, rather than a generic version of that category of product, are still unreliable: proportions, logos, packaging text and fine detail frequently come out wrong, which is exactly the detail a customer will notice.
Hands, eyes, teeth and the small, involuntary movements of a real face in video are the areas current generation tools are weakest on, and a Dubai audience used to seeing genuine people in ads notices the artificial version quickly.
A model trained mostly on global content tends to produce Arabic copy or imagery that reads as generically translated rather than written for a Gulf audience, and it will not reliably catch a phrase or an image that reads as tone deaf locally.
A language model will confidently write a specific claim about your product, a price point, a comparison to a competitor or a statistic, with no reliable way to know whether it is true, so every factual claim it produces has to be checked against what you can actually stand behind.
The same generated character or product can look subtly different from one image to the next unless a person is actively steering and correcting for it, which undermines the polished, consistent look a real campaign needs.
AI tools can produce more variants, but nothing about the generation process tells you in advance which one a Dubai audience will respond to. That is still decided by testing in the account, not by the tool that made the asset.
What still needs a person
None of the limits above mean AI has no place in Dubai ad creative. They mean it sits inside a process a person still runs, not the other way round. A person sets the brief and the claims that are allowed to be made. A person selects which generated drafts are strong enough to refine rather than publishing the first output. A person checks every generated image or video against the product it claims to show, and rejects anything that misrepresents it. A person decides what the final creative says, what it promises, and whether it complies with the platform and the UAE rules that apply to it. That division of labour is the whole point of how we run AI ads for Dubai clients.
We treat AI the same way we would treat a very fast junior designer: useful for producing options quickly, not trusted to make the final call on what represents your business publicly. Every AI assisted creative we deliver has gone through a human edit and a human sign off before it is called finished, whether the brief was for a single banner or a full set of AI ads for a Dubai launch.
Our process
The same six stage way of working sits underneath this like every other service, with the AI specific checkpoints added where they actually happen.
What the ad needs to say, what it is allowed to claim, and which parts of the brief are candidates for AI assisted drafting versus needing a real shoot or a licensed asset instead.
A fixed written quote scoped to your brief, setting out which parts of the work use AI assisted tools and which do not, sent within 45 minutes during business hours.
An agreed number of directions, produced with AI assisted drafting where it is suitable, reviewed by a person before anything is shown to you as a serious option.
Every surviving concept is edited, corrected against your actual product and claims, and checked for anything that misrepresents what you offer.
Approved creative is built out into the aspect ratios and languages the campaign needs, with a native reader checking any Arabic copy before it is finalised.
Each finished asset is checked against the current Meta, Google and TikTok rules for AI generated or edited content, and labelled or flagged for disclosure where the platform requires it.
What we need from you
The quality of AI assisted output depends heavily on what it is given to work from, more than most clients expect.
Even where a background or scene is generated, an accurate depiction of your actual product usually needs to be built from real photography of it, not generated from a text description alone.
The specific facts, figures and comparisons you are willing to defend if challenged, so generated copy has real boundaries rather than inventing plausible sounding claims on your behalf.
Signed consent from any employee or customer whose face, voice or likeness appears or is recreated, covering the specific use, before any generation work involving them starts.
Existing ads, competitor examples you like or dislike, and your current brand guidelines, so AI assisted drafts start closer to your actual voice rather than a generic default.
AI assisted drafting speeds up early concepts, but human review, fact checking and compliance checking still take real time and should not be compressed out of the schedule.
A single person, or a small named group, with authority to approve what is allowed to run, since a generated ad going live without a clear sign off is the fastest way to publish something nobody actually checked.
Platform rules
Each of the three main platforms now has a published rule about AI generated or digitally altered content in advertising, and they are not identical. Meta’s advertising standards require disclosure where a photorealistic image, video or audio in an ad was created or edited using third party generative AI tools to depict a real person saying or doing something they did not do, a realistic person or event that is not real, or an alteration of authentic footage, and Meta states it also runs automated detection on top of advertiser self disclosure, applying an AI info label to what it identifies, which is the first rule any Dubai advertiser running AI ads on Facebook or Instagram needs to know. That standard is strictest for ads about social issues, elections or politics, where verified disclaimers and an authorisation process apply on top of the AI disclosure itself.
Google’s current approach, in force from July 2026, lets advertisers add a text or visual AI label directly on image and video creative, or switch on an AI label setting so Google discloses it automatically in the ad’s “how this ad was made” panel, specifically to meet transparency rules already in force in the European Union, India and New York. Election advertising carries a separate, older requirement to disclose synthetic or digitally altered content through a dedicated checkbox in campaign settings. TikTok requires advertisers to check a toggle stating the ad contains AI generated content whenever the creative is completely AI generated or significantly modified by AI, and once submitted that disclosure cannot be quietly removed without changing the underlying asset. We check the live rule on each platform against the specific creative before any AI ads for a Dubai account go out under our name.
| Platform | What triggers disclosure | How it is disclosed |
|---|---|---|
| Meta | Photorealistic AI created or edited image, video or audio showing a real person doing or saying something they did not, or a realistic person or event that is not real | Advertiser self disclosure at ad creation, plus an automated AI info label Meta applies where it detects it |
| Image or video assets generated or substantially modified with generative AI, with stricter rules for election ads with synthetic or altered content | A visible label on the creative or the AI label setting, shown in the ad’s “how this ad was made” panel | |
| TikTok | Image, video or audio that is completely AI generated, or significantly modified by AI | An “AI generated content” toggle in Ads Manager, set before the ad is submitted |
Summarised from each platform’s own published advertising policy pages, read on 25 September 2026. Platform policy changes frequently; we check the live version against your specific creative before it is submitted, and this is not legal advice.
UAE rules
Advertising content shared for compensation or free of charge on social media or digital platforms in the UAE sits under the National Media Authority’s permit framework regardless of whether AI was involved in making it, so an AI generated ad still needs the same advertiser permit consideration as one shot conventionally, which our guide to the UAE advertiser permit covers in full, and it is the reason AI ads for a Dubai account still need the same permit thinking as a conventionally produced one. Separately, and specifically about AI, the UAE Media Council has publicly warned that using artificial intelligence to depict public figures or national symbols without prior official approval is a clear violation of media content standards, and that using AI to spread misinformation, incite hate speech or defame someone is treated as a media offence with fines and administrative penalties attached. That means an AI generated ad involving a recognisable public figure, a government symbol or a national emblem needs specific care and, in most cases, should not be attempted without prior approval.
Where an AI generated asset is displayed publicly as an advertisement, the underlying Dubai advertising permit and Arabic content rules covered on our graphic design page still apply on top of the platform specific disclosure rules above; being AI generated does not exempt a piece of advertising from the ordinary rules that already govern advertising in Dubai. AI ads in Dubai are still advertisements first, and generated second.
Consent
This is the point we will not soften: it needs written consent, every time.
Creating a synthetic likeness of a real, identifiable person, whether that is an employee, a customer or anyone else, and using it in an advertisement without their agreement, sits inside UAE privacy law, not outside it because AI was involved in producing it. Federal Decree-Law No. 34 of 2021 on Countering Rumours and Cybercrimes addresses taking, retaining or publishing a person’s image or voice recording without consent, and its penalties increase specifically where a recording, image or scene has been modified or altered. Our guide to photography and consent under UAE law covers the detail; the point that matters for AI ads is that generating a synthetic version of someone’s face or voice is a modification of their likeness in exactly the sense that law is concerned with, and it needs their written agreement before it happens, not after. This single rule stops more AI ads projects in Dubai at the brief stage than any platform policy does.
In practice this means we ask for signed, specific consent before we build an AI generated likeness of any real person, whether an employee representing your brand or a customer whose testimonial you want to recreate, and we keep that consent on file as part of the project, whether the piece is a single AI ad or a full set of AI ads running across a Dubai campaign.
UAE copyright law, set out in Federal Decree-Law No. 38 of 2021, was written around a human author, and legal commentary on it is consistent that autonomous AI output with no meaningful human contribution is not likely to automatically qualify for the same protection as a piece a person made directly, because rights under the law attach to a creator rather than to a tool. That leaves the honest position unsettled rather than simple: where an ad creative has substantial human direction, editing and selection behind it, that human contribution is the stronger basis for ownership, and we document it as part of the project rather than treating the generated output alone as automatically and fully yours by default.
What we do commit to is handing over the working files, prompts and edit history for anything we produce, so your position is as strong as it can currently be, and we say plainly when a specific piece of output sits in a genuinely untested area of the law rather than promising a certainty that does not yet exist, a caveat that applies to every set of AI ads we produce for a Dubai client.
Mistakes we see
Almost all of these come from treating the output as finished rather than as a draft.
A generated image or line of copy published without a person checking it against the actual product and claims, which is where product inaccuracies and overreaching claims most often slip through.
A realistic AI generated scene or voice running without the disclosure the platform’s own policy required, which risks rejection, a policy strike, or worse, a viewer feeling deliberately misled.
An employee’s face or voice recreated for a campaign because it was convenient, without a signed agreement covering that specific use, which is a privacy problem regardless of how good the result looks.
A translation that reads as technically correct but tonally off being published because it “looked fine”, when a native reader would have caught the phrasing in seconds.
Spending media budget testing a large batch of near identical AI generated variants instead of a person first narrowing them down to the handful genuinely worth a customer’s attention.
Underestimating the human review, fact checking and compliance time AI assisted creative still needs, and being surprised the total cost did not fall as much as expected.
Judging the result
The measure is not whether it looks impressive as a generated image. It is whether it did what an ad is for.
| What was made | What tells you it worked |
|---|---|
| A batch of AI assisted copy variants | Whether the strongest lines survive a human edit and read as something your business would actually say |
| A generated concept visual | Whether it accurately represents what you sell, not just whether it looks polished on screen |
| A resized and localised creative set | Whether every format and language version reads correctly, with nothing cropped, mistranslated or left disclosed incorrectly |
| The finished ad in market | Whether it performs against your own campaign goals, measured the same way any other ad creative is measured, not against a separate AI benchmark |
We do not report invented performance figures for AI generated creative against conventionally produced creative, because we have not measured a reliable, general pattern either way, and any agency claiming one without your own account data to back it up is guessing. That honesty is worth more to a Dubai business weighing up AI ads than a confident but unverifiable number. What we commit to is creative that has been through the process above before it is called finished, and a plain explanation of which parts used AI once it is live.
Dynamic creative
Several platforms now offer some form of dynamic creative optimisation, where a system automatically combines approved headlines, images and calls to action into many permutations and shows more of whichever combination performs best. This is genuinely useful, and it predates the current wave of generative AI by several years, even though it is often sold under the same banner now. The part worth understanding is what the system is actually doing: it is not writing new copy or generating new images on its own, it is recombining assets a person already approved, and testing which combination a given audience responds to.
That distinction matters for a Dubai campaign because it changes what quality control looks like. Every individual asset that could appear in a dynamic combination has to be checked before it goes into the pool, since the system will happily combine a strong headline with a weak image if both are available to it, and it has no sense of which pairings look coherent as a finished ad, which is exactly why dynamic AI ads for a Dubai account still need a person curating the pool. We build the pool of approved assets first, with the same human review as anything else on this page, then let the platform’s own optimisation decide which combinations to show more of. That is how dynamic, partly AI assisted ads stay reliable for a Dubai account rather than drifting off brand.
Where AI fits by format
The role AI plays is not the same for a search text ad as it is for a video placement, and treating them identically produces weaker results in both.
| Ad format | What AI assisted tools help with here | What still needs a person or real asset |
|---|---|---|
| Search text ads | Generating headline and description variants within a platform’s character limits | Choosing which claims are accurate and compliant to actually submit |
| Display and banner ads | Concept visuals, background generation, resizing across the many banner sizes a campaign needs | An accurate depiction of the real product where one appears |
| Social feed image ads | Fast concept exploration, copy variants matched to platform tone | Final selection and a human check against the brand and the offer |
| Video and Reels style ads | Storyboarding, voiceover script drafts, some background and motion elements | Anything showing a real product closely or a real person; see AI video production |
| Shopping and catalogue ads | Copy variants for product titles and descriptions within feed limits | Accurate product data pulled from your actual catalogue, not generated |
A short list of practical limits is more useful here than a long statement of principle. We will not present AI generated content as a genuine customer testimonial, because that misrepresents who is speaking and misleads the person watching it, which is the same reasoning covered in full on our AI UGC page. We will not generate a synthetic likeness of a real, identifiable person without their written consent on file first, whoever they are. We will not let a generated claim about your product, a comparison, a statistic or a result, go into an ad without checking it against something you can actually stand behind, since a confident sounding claim is not the same thing as a true one. We will not skip the disclosure step a platform’s current policy requires because it is inconvenient or because a generated asset is borderline; where it is genuinely unclear, we treat it as requiring disclosure rather than assuming it does not.
None of these limits are about being cautious for its own sake. They are the difference between AI assisted creative that survives a platform review, a regulator’s attention and a customer’s trust, and creative that looks fine until one of those three notices something was not disclosed, was not consented to, or was not true.
Testing before scaling
The same discipline that applies to any new ad creative applies here, and skipping it is the most common way an AI assisted campaign disappoints.
An agreed, limited set of AI assisted concepts tested alongside your existing creative, not a wholesale replacement of everything running, so you have a fair comparison rather than a leap of faith.
AI assisted creative is judged against the same metrics as everything else in the account, not a separate, softer standard invented to make it look better.
Before scaling up whatever performed best, a person checks it is not winning for a reason that will not hold up, such as a claim that will not survive scrutiny or a look that will tire quickly.
Only once a batch has genuinely earned it does more budget and more variants of that direction get produced, rather than assuming an early result will keep repeating indefinitely.
Arabic copy in practice
Asked to translate an English ad line into Arabic, a language model will usually produce something grammatically correct and technically readable, which is exactly why the mistakes it makes are dangerous: they do not look like mistakes on a first read. Idioms translate literally and lose their meaning. A tone that reads as confident and punchy in English can read as blunt or oddly formal in Arabic. Product names or slogans that were never meant to be translated sometimes get translated anyway. None of this shows up in a spell check, because nothing is spelled wrong, it simply does not read the way a native Gulf Arabic speaker would naturally write it.
We treat an AI generated Arabic draft as exactly that, a draft, and a native Arabic reader who understands Dubai advertising specifically reviews it before it is submitted to any platform, checking meaning and tone against the English original rather than checking grammar alone. This is the same standard we apply to Arabic and English design work generally, and it does not change because the first draft came from a model instead of a person; bilingual AI ads for a Dubai audience only work once a native reader has signed off the Arabic.
Reporting
The report should tell you what ran and what it did, not how advanced the tools sound.
A plain note on which headlines, images or video elements were AI assisted and at what stage, so you always know what you are looking at rather than guessing.
The same metrics used for every other ad in your account, so an AI assisted variant is judged on the same basis as one that was not, with no separate scale invented to flatter it.
A clear account of what was dropped, what was kept and what is being tried next, so a testing programme reads as a deliberate process rather than a stream of unconnected variants.
This page covers the creative itself and the rules around making it with AI. The account structure, targeting, budget and bidding strategy that the creative runs inside sit on PPC advertising for search and shopping, Google Ads management for the full Google account, and social media advertising for Meta, TikTok and the other social platforms. Where the brief is about short, creator style clips made at volume specifically for paid social testing, that is a distinct job covered on AI UGC, and where the goal is a finished piece of video as the deliverable itself, such as a product or explainer video, that sits on AI video production. Wider campaign planning across all of these sits on marketing strategy, and every service on this page belongs to the digital marketing services hub alongside the others. Wherever your brief sits on that list, the same honest approach to AI ads for Dubai audiences applies.
If a campaign genuinely calls for a paid creator’s own following and credibility rather than an AI generated persona, that is a different service again, covered on influencer marketing, and the same advertiser permit and disclosure logic covered above applies there too, whether the campaign is built around AI ads, a Dubai based creator, or a mix of both.
From the blog

Who needs a UAE advertiser permit, how the visitor permit differs, and the advert rules brands in Dubai must follow when they pay creators.
Read the article
What UAE law and TDRA policy actually require before a business in Dubai sends a promotional text, and the records you should keep.
Read the article
Opt in, templates, messaging limits and UAE law: what a Dubai business needs to know before sending promotions on WhatsApp.
Read the articleStraight answers
Often yes. Meta requires disclosure when a photorealistic image, video or audio in an ad was created or edited with generative AI to depict a real person doing something they did not do, or a realistic person or event that is not real, and it applies automated detection on top of self disclosure. Google's AI label setting and TikTok's AI disclosure toggle cover a similar range of content. We check each platform's current rule against the specific creative before it runs.
Only with their written consent. UAE law treats using someone's image or voice, including an AI altered version of it, without their agreement as a privacy matter, and Meta, Google and TikTok all require consent for a synthetic likeness of a real, identifiable person. We ask for that consent as part of scoping, in writing, before any generation work starts.
UAE copyright law was written around human authorship, and it does not yet directly answer who owns output with no meaningful human input. We treat this as unsettled rather than promising outright ownership, document the human direction, editing and selection that went into the finished creative, and hand over the working files and prompts so the position is as clear as it can currently be.
We do not promise that, because it depends entirely on the product, the audience and the platform, and we have not measured a reliable pattern either way. What AI does reliably help with is producing more concepts, more format variants and faster first drafts to test, which a human then edits and a human still decides what actually runs.
Some of it, and we say plainly which parts. AI helps with copy variants, first draft visuals, resizing across formats and some editing steps. Targeting decisions, budget allocation, bidding strategy, compliance sign off and what actually gets published are handled by a person, and we can explain exactly which of your ad's assets used AI and which did not.
We do not quote a fixed saving, because the honest answer depends on the brief. What AI genuinely removes is some of the time spent on early drafts and format variants. A fixed written quote scoped to your specific brief, including which parts use AI, is the only reliable way to know the cost for your project.
Sources
Fixed price, in writing
Got it. Your quote is being written now.
In business hours you will have it within 45 minutes. Check your inbox for the confirmation.