AI Video Rules and Labelling in the UAE

The evidence behind that opening answer is worth showing rather than asserting, because it is the kind of claim a reader should be able to check.

Google's own advertising policy, published as it rolls out AI labelling from July 2026, explains what is driving the requirement. It names the regulations that oblige advertisers to disclose AI-generated or edited assets: the European Union, India, and New York. The UAE is not among them.

That is a useful signal and it should be read for exactly what it is. It shows which jurisdictions have written AI advertising rules that a global platform has had to build for. It does not show that the UAE has considered the question and decided against, and it is not a substitute for asking the Council. A brand about to spend seriously on generative creative should ask directly and file the reply.

Which brings the immediate correction, because the wrong takeaway from all of this is the dangerous one. No labelling rule does not mean no rules. Three layers govern a film containing AI in this market, and only one of them is about AI at all.

The first is the content standards, which apply to advertising regardless of how it was produced. The second is the platform, which is enforcing today and enforcing automatically. The third is the contract, meaning what the client owns and what the people on screen consented to, where the honest answer is currently incomplete.

The Journal's article on the UAE advertiser permit is the companion piece to this one, and it covers the standards regime this sits inside.

For most brands, the binding obligation today comes from where the film runs rather than from where it was made.

Google states that starting in July 2026, advertisers can add text or visual labels directly within image and video ad creatives that were generated or modified using AI, with an AI label setting rolling out gradually through Google Ads, Display and Video 360, Campaign Manager 360, Merchant Center and Ads Editor.

One sentence on that page matters more than the feature announcement, and it deserves reading twice. Google cautions that use of the AI label setting in its advertising products does not guarantee compliance with specific regulations, and advises advertisers to seek legal guidance and take measures as needed to ensure their ads and assets follow all local legal obligations and policies. The label is a tool. It is not a shield, and a brand that treats a toggle in an ad platform as a compliance position has misunderstood what it bought.

Google handles election advertising separately, with its own requirement to disclose synthetic or digitally altered content through campaign settings.

Meta's current position comes from industry commentary rather than from its policy text, which was not retrieved for this article, so it is reported here as commentary. That commentary describes Meta as having moved AI disclosure in advertising from optional to enforced during 2026, with sponsored content using AI-generated visuals, text or audio required to carry disclosure on Facebook and Instagram. Anyone planning paid distribution should read Meta's own current policy rather than relying on any summary of it, including this one.

The production consequence is concrete. The platform a film will run on is part of its specification, in the same way an aspect ratio is. A film containing generative elements built for an organic post and the same film built for paid distribution are governed differently, and that difference belongs in the brief rather than in a conversation after the media plan lands.

This Journal has now found the same shape twice. On music, and now on AI, platform policy binds earlier than statute, enforces faster, and enforces automatically, with no notice and no appeal worth the name.

The sharpest risk in generative work is not the skyline. It is the person.

Article 16 of the same law gives performers moral rights that are perpetual and that may not prescribe or be assigned. Those rights entitle a performer to attribute the performance to themselves, and to prohibit alteration, distortion, defacement or modification of their performance in a manner that would prejudice their reputation.

Set that against what generative tools now do routinely to recorded material: extending a take, altering a delivery, cleaning up a line, changing what a person appears to say or do. On its face, that is the activity the provision contemplates. Whether a particular edit crosses the threshold was not established by this research, and the threshold itself is unknown, so the useful position is caution rather than a rule.

The drafting answer is the one the Journal's casting article already reaches, updated for the tools. A talent agreement should address re-editing, re-versioning and use in contexts other than the one filmed, and in 2026 it should name generative alteration explicitly rather than relying on general language written before the capability existed. The moral right itself cannot be signed away, so the clause is about defining scope and expectation rather than about obtaining a waiver.

Two things sit alongside that. The Media Council's prohibition on depicting public figures without official approval reaches people the production never filmed at all, which is a category that did not previously need managing. And the privacy layer covered in the Journal's article on filming people applies to any real person used as source material, because nothing about a generative process removes the consent that would have been needed to film them.

One open question is worth naming rather than answering. How the public-figure prohibition applies to a synthetic person who resembles nobody in particular was not established, and it is the question the market will test first.

Decide at treatment stage which elements, if any, will be generative, and write the decision down. Retrofitting that question after an edit has been approved is exactly how it gets missed, and it is also how a client finds out in a review rather than in a briefing.

Ask the platform question before the creative question. Where does this run, and does that platform require a label. Today that is a more consequential question than any statutory one.

Ask the Media Council directly about any UAE disclosure expectation, rather than relying on this article or any other, and keep the written answer on file. The finding here is that no such requirement was found. That is not the same as a guarantee, and it should not be treated as one.

Put the generative elements into the delivery documentation. Update talent agreements to name generative alteration. Apply the buyable test at briefing rather than at delivery.

And watch the Federal Authority for Artificial Intelligence and Data, whose mandate includes proposing legislation. The current position is a snapshot of August 2026, not a settled destination.

On cost, indicative ranges only. A solo videographer sits around AED 3,500, the fast-turn end of the market and work Grolez Films does not take. A DOP and small crew runs roughly AED 25,000 to 45,000. A full production starts from AED 100,000. Generative tools change the composition of a budget rather than its tier, and they do nothing at all to the costs that actually dominate production in this market, which are access, permits, consent and time. That is the honest correction to the savings pitch clients are currently being given.

This article makes no claim about generative work at Grolez Films, in either direction, and the Journal is deliberately silent on the subject while the record is thin. Nothing here should be read as an offer to do any of it. What is on the record is the filmed job.

Luis Grolez, the founder, pitches this personally, founder to founder. Get in touch about a specific brief and a specific quarter.