By THE UNKNOWN BRAND · Editorial standards

Published 2026-08-01 · Updated 2026-09-06

The strongest AI-production advice sometimes ends with “book the camera.” Use the method that protects the idea and the audience’s trust.

01Kuwait Army02Gulf Bank03Midana04Warba Bank05Dabdoob06Boursa Kuwait07Palmy08World Defense Show09Hyundai10Kellyville

In brief

• YouTube, Meta and TikTok all require disclosure or labelling when realistic content is generated or altered, and regulated categories add approvals on top.
• A disclosure proves the audience was told the content was generated. It does not prove the claim inside the film is accurate.
• The international marketing code makes the point that audiovisual material misleading about a product’s characteristics or performance should not be used, generated material included.
• Neither method is automatically cheaper. AI-led production tends to move cost from logistics, crew and location to supervision, reference preparation and quality control.
• Where a film’s whole value is that something happened, the answer is a crew. A generated reconstruction is not a substitute for proof.

Choose a physical shoot for evidence

Film the real event, testimony, place, spokesperson, craft process or product behaviour when its truth is the reason the audience should believe it. A generated reconstruction may be useful as illustration, but it is not a substitute for proof, and presenting it as proof spends trust the brand will want later.

Evidence is the real event, testimony, place, spokesperson, craft process or product behaviour whose truth is the reason the audience should believe it. A generated reconstruction may be useful as illustration, but it is not a substitute for proof.

Choose AI-led production for controlled invention

AI is strong when the world is intentionally impossible, versions are numerous, continuity can be governed from reference material, and the work is clearly advertising or designed communication rather than documentary fact. The audience is not being asked to believe the scene happened; it is being asked to enjoy what the scene means. That framing is what keeps invention honest: the more clearly a film reads as designed, the more freedom the design can take.

AI-led production is generative work for a world that is intentionally impossible, versions that are numerous, and continuity that can be governed from reference material. It is strong where the work is clearly advertising or designed communication rather than documentary fact.

Choose hybrid when the product must be exact

Capture the product, person or location accurately, then use generative production for environments, transitions, extensions or versioning. Hybrid often delivers the best balance of evidence and freedom.

Hybrid production is capturing the product, person or location accurately, then using generative production for environments, transitions, extensions or versioning. It often delivers the best balance of evidence and freedom.

The briefs a generative studio should refuse

A studio that only sells generative production has an obvious incentive to find AI in every brief. The honest position is narrower.

Where a film’s whole value is that something happened, the answer is a crew. The ribbon was cut. The machine ran. The customer said the sentence. A generated version documents none of that, however convincing it looks, and convincing is rarely the hard part.

The international marketing code makes the same point from the other side. Audiovisual material that misleads about a product’s characteristics or performance should not be used, including where it misleads because it was generated.

THE UNKNOWN BRAND says this before a treatment is written rather than after. It costs a project occasionally. It costs less than a client meeting the limit at an approval gate with the shoot date already gone.

A decision table by brief type

The verdicts below are defaults, not rules. Any row can be overturned by the risk questions at the end of this guide, and several briefs will sit between rows. The table exists so the conversation starts from the nature of the brief rather than from a preference for one production method.

Use the table at the first conversation, not the last. When a brief matches two rows, price both routes and let the risk questions decide; when it matches none, write down which parts of the film are evidence and which are invention, and the verdict usually writes itself.

Cost and schedule honesty

Neither method is automatically cheaper. AI-led production tends to move cost from logistics, crew and location to supervision, reference preparation and quality control; a physical shoot spends more on the day and less on review. As general industry experience rather than a promise: the briefs where AI saves most are those needing many versions of a governed idea, and the briefs where it saves least are those needing one perfect, exact rendition of a real thing. Schedules follow the same logic, and a hybrid plan should budget for both kinds of work rather than assuming the generative half is free.

Decide disclosure before production, not after

The large platforms have already settled this in writing. The wording differs; the substance agrees. Realistic generated content gets labelled.

• YouTube: disclosure is required when content makes a real person appear to say or do something they did not, alters footage of a real event or place, or generates a realistic scene that did not occur.
• Meta: people must disclose organic posts carrying photorealistic video or realistic-sounding audio that was digitally created or altered.
• TikTok: AI-generated content containing realistic images, audio or video must be labelled.

Regulated categories add their own approvals on top of that. Decide the disclosure at the brief, because it is a design problem before it is a legal one. A label chosen early is part of the film. A label added after the master is signed off reopens the sign-off.

What a disclosure settles is narrow. It proves the audience was told the content was generated. It does not prove the claim inside the film is accurate, that the likeness or location was licensed, or that the product behaves as shown.

Disclosure is the label telling an audience that content was generated or altered. What it settles is narrow. It proves the audience was told. It does not prove the claim inside the film is accurate, that the likeness or location was licensed, or that the product behaves as shown.

Risk questions

Before the method is final, answer these in writing:

• Could a reasonable viewer misunderstand what is real?
• Does a regulator, platform or client policy require disclosure?
• Is the likeness, voice, logo, location or style licensed?
• Would an error change a purchase, health, financial or public-service decision?
• Can the team reproduce the approved identity across every final asset?

The strongest AI-production advice sometimes ends with “book the camera”. Use the method that protects the idea and the audience’s trust, because no production saving is worth what a misleading frame costs.

Risk questions are the checks answered in writing before the method is final. They cover what a reasonable viewer could misunderstand, whether a regulator, platform or client policy requires disclosure, whether the likeness, voice, logo, location or style is licensed, and whether an error would change a purchase, health, financial or public-service decision.

Sources

Every source below was opened and its quoted wording checked against the live page. A source we could not re-fetch was dropped, not softened. Where a rule could not be confirmed from the body that issues it, this guide says so rather than describe it from memory. Editorial responsibility sits with the studio, not an individual author. Found a moved link or a wrong citation? Email hello@theunknownbrand.com with the URL. We will correct the page and its modification date.

1. YouTube Help (Google) — The post’s risk question 'Does a regulator, platform or client policy require disclosure?' and the table verdict that a scene the audience would reasonably feel misled by must be handled differently. Also cited in the new disclosure section.
2. Meta Transparency Center — Same disclosure risk question, from a second platform. Supports the new section’s claim that the major platforms have already settled this in writing.
3. TikTok Newsroom — Third platform for the same disclosure point, and the specific wording that the trigger is realism rather than the use of AI as such.
4. International Chamber of Commerce (ICC Advertising and Marketing Communications Code, September 2024 edition) — Three separate existing claims: (1) that a generated reconstruction is not a substitute for proof and that a viewer misled by a generated scene is the failure case; (2) the table verdict sending testimonial and interview films to a physical shoot; (3) the risk question 'Is the likeness, voice, logo, location or style licensed?'. Note: WebFetch downloaded the PDF but could not parse it, so I extracted and read the text locally with pdftotext from the file WebFetch saved in this task.

Frequently asked questions

التصوير الحقيقي أم الصور المولَّدة: إطار قرار قبل إنتاج أي صورة — هذا الدليل بالعربية.