For legal, procurement and brand-safety teams
AI ad governance: rights, labels and product accuracy.
AI Vidia (Vidia ApS, CVR 44873443, Denmark) makes AI video and image ads for consumer brands. This page is for the legal, procurement and brand-safety teams who check a studio before a pilot.
It sets out how a job runs, how AI content is labelled, how faces and voices are handled, how product accuracy is checked, who owns what, and which points are agreed in the contract instead.
Last reviewed
Process
How a job runs.
- 01
Product facts first
You send the product photos you already have and sign a written product fact sheet. Claims in the ads come only from that sheet.
- 02
Approve the look
You approve the look before anything is generated.
- 03
Review every file
Every file passes AI review checks and is checked against photos of your real product. On client work, a creative director checks every file. The first review batch arrives on day 3.
- 04
One revision round
Each video includes one revision round. Brief to delivery takes 7 days.
- 05
Approve and launch
You approve the final files, each with a usage note listing what is AI-generated. Your team or ours launches and tests them, as agreed per client.
- 9:16 is the standard format; other formats can be agreed per client on request.
- Existing footage you send can be cut into the weekly batches.
- English and Danish are reviewed in-house. Other languages go to your own reviewer before anything ships.
Disclosure
AI labels and disclosure.
- Each delivered file comes with a usage note that lists what in it is AI-generated.
- We agree the AI label with you per platform and market before delivery.
- Approval is each platform's decision: no producer can guarantee it.
The rules on 21 October 2026
A dated summary of the rules brand teams ask us about most, in the markets we sell into. Each row links its source with the source's date. It is a starting point for your counsel, not legal advice.
| Rule | Status | What it asks of an ad | Sources |
|---|---|---|---|
| European Union, including IrelandAI Act, Article 50(4) | Applies since 2 August 2026 | Realistic AI-generated or AI-altered images, video or audio that could pass as real must be disclosed clearly, at the latest at first exposure. The Commission's July 2026 guidelines give realistic AI presenters and AI product images that change how a product looks as examples. Trade guidance from IAB Austria (September 2026) reads the agency as the usual party that labels. |
|
| New York, United StatesGeneral Business Law section 396-b | In force since 9 June 2026 | An ad seen in New York that includes a synthetic performer, meaning a realistic AI-made person who is not a recognisable real performer, needs a conspicuous disclosure when the advertiser knows. Fines are USD 1,000 for a first violation and USD 5,000 for each later one. |
|
| California, United StatesSB 1050 | Signed 16 September 2026; applies from 1 January 2027 | A similar disclosure when an ad prominently includes a synthetic performer, AI voices included. The law gives safe-harbour wording, such as "this performance features a synthetic performer". |
|
| United States, federalFTC Rule on Consumer Reviews and Testimonials | In force since 21 October 2024 | Bans fake reviews and testimonials, including ones by people who do not exist, such as AI-generated reviewers, or by people with no real experience of the product. |
|
| United KingdomCAP and BCAP Codes, ASA guidance | In force; ASA guidance on AI and deepfakes published 11 June 2026 | No general legal duty to label AI. The test is whether the ad misleads, and the advertiser stays responsible for any AI output it uses. A disclosure is very unlikely to fix an AI image that shows results the product does not deliver. |
|
| CanadaCompetition Act; Ad Standards disclosure guidelines | In force; guidelines updated fall 2025 | No AI-labelling law. The Competition Bureau says there is no "AI exception" to deceptive-marketing law. Ad Standards asks for a disclosure, such as #MadeWithAI, when an image showing product benefits is AI-generated or significantly AI-altered. |
|
| Australia and New ZealandAustralian Consumer Law and AANA Code; NZ Advertising Standards Code | In force; New Zealand's updated Code applies to new ads from 1 February 2027 | No AI-labelling law. Ad Standards Australia says the advertising rules apply however the content was made. New Zealand's code already covers AI ads, and new ASA guidance says to consider a disclosure when AI use could change how people read an ad, such as a realistic person who does not exist; a label alone does not stop an ad misleading. |
|
Platform labels
| Rule | Status | What it asks of an ad | Sources |
|---|---|---|---|
| TikTokAds policy: misleading and false content | Updated April 2026 | AI-generated or significantly edited ad content needs TikTok's AI-generated content label or a clear disclaimer, caption, watermark or sticker of your own. Without one, the ad can be rejected or restricted. |
|
| Meta: Facebook and InstagramAI transparency for ads | Updated 1 June 2026 | Meta labels ads made or significantly edited with its own generative AI tools, next to "Sponsored" when they show a photorealistic AI person, and says it will detect ads made with third-party AI tools through industry-standard signals, labelled "AI info" under "About this ad". |
|
| Google Ads, including YouTube adsAI labelling update | Posted 9 July 2026 | Google permits AI labels inside image and video ads and offers an AI label setting, while noting that the setting does not guarantee compliance with local law. Election ads must declare altered or synthetic content. |
|
Laws, guidance and platform policies change, and how they apply depends on your campaign and its markets.
For the full picture, read AI labels on ads in 2026: what Meta, Google, TikTok and the EU require.
People
Faces, voices and likeness.
- AI presenters are AI-generated, not customers. They deliver scripts you approve, as spokespeople for your brand.
- Product claims in a script come only from your signed fact sheet.
- A real person's face or voice is used only with their written consent and an agreed use.
- The table above lists the US rules on fake testimonials and on synthetic performers.
Accuracy
Product accuracy and claims.
- Claims come only from the product fact sheet you sign before anything is generated.
- Every file is checked against photos of your real product: AI review passes first, then, on client work, a creative director checks every file.
- If an ad shows the product in a state your photos do not, such as open, tell us before production and send a photo of that state if you have one.
- A shot offered as proof that the product works should be real footage: a generated demonstration is an illustration, not proof.
- Demo work on this site uses fictional brands, such as HAVREL and OSTRID, labelled as demos.
How a brand team can review AI ads before sign-off: Reviewing AI ads before approval.
Rights
Who owns what.
You own the finished files from delivery. The signed scope of work sets out the details.
The product photos and brand assets you send us stay yours.
What an agreement on AI ad rights should cover: AI ad usage rights: what the agreement should clarify.
Contract
What is agreed in the contract.
This page describes how we work; it does not set contract terms. The points below are agreed in the written agreement with each client.
- Scope, volumes and the terms of the engagement
- The details of ownership and use of the finished files
- Confidentiality
- The AI tools used on your work
- Who applies the AI label in each market
- Who launches and tests the ads: your team or ours
You can raise any of them with ken@aividia.dk before a pilot.
Questions
What legal and procurement teams ask.
Who applies the AI label to our ads: you or us?
We agree the AI label with you for each platform and market before delivery, and each file's usage note lists what in it is AI-generated. For the EU AI Act, trade guidance from IAB Austria (September 2026) reads the agency as the usual party that labels.
Who applies the label on your campaigns is set in the written agreement.
Are your AI presenters real people or real customers?
No. AI presenters are AI-generated, not customers, and they deliver scripts you approve. Product claims come only from your signed fact sheet. A real person's face or voice is used only with their written consent and an agreed use.
How do you keep AI from changing how our product looks?
Every file is checked against photos of your real product: AI review passes first, then, on client work, a creative director checks every file. Claims come only from the product fact sheet you sign before anything is generated.
If an ad shows the product in a state your photos do not, such as open, tell us before production.
Who owns the finished ads and the files we send you?
You own the finished files from delivery. The signed scope of work sets out the details. The product photos and brand assets you send us stay yours.
Which AI tools will be used on our work?
The AI tools used on your work are agreed in the written agreement, with the other points listed under "What is agreed in the contract" on this page. You can raise them with ken@aividia.dk before a pilot.
How does our legal review fit into production?
You approve the look before anything is generated and the final files before launch. The first review batch arrives on day 3, brief to delivery takes 7 days, and each video includes one revision round.
We review English and Danish in-house; other languages go to your own reviewer before anything ships.
Can you guarantee our AI ads pass Meta, TikTok and Google review?
No. Approval is each platform's decision, and no producer can guarantee it. Each file's usage note lists what is AI-generated, and the AI label is agreed per platform and market before delivery, so each platform's AI disclosure can be set from a written record.
Related pages
Company facts
Vidia ApS, CVR 44873443, founded 2024: the team, the office address and official links.
AI Vidia company informationFor brand teams
What working with AI Vidia looks like for an in-house brand team.
AI Vidia for brand teamsFor agencies
How an agency can add AI ad production for its clients.
AI Vidia for agency partnersPlans and contracts
Four published plans in EUR, from the EUR 2,000 Control Test to the EUR 18,000 a month Brand System, and how contracts and invoicing work. Media spend is not included.
AI Vidia plans and pricesQuestions from legal or procurement?
Send them before a pilot. Points that depend on your contract are settled in the written agreement, not on this page.
ken@aividia.dk