Are AI-Generated UGC Ads Legal in Europe?

Artificial intelligence has changed the way brands produce advertising content. Businesses can now create virtual presenters, synthetic voices, product demonstrations and social media videos in a fraction of the time previously required.

Tools such as Higgsfield, Veo, Runway, Kling and Synthesia have made this type of production faster and more affordable. However, they have also created an important question for brands, agencies and online retailers: are AI-generated UGC adverts legal in Europe?

The simple answer is that AI-generated UGC adverts are not generally banned in the European Union. However, certain realistic AI-generated or manipulated content must be clearly disclosed. It is also important to understand that adding an "AI-generated" label does not automatically make an advert lawful. A misleading product claim, fabricated customer testimonial or unauthorised use of someone's face or voice may still break other laws.

This guide explains the transparency rules that apply from 2 August 2026, when brands may need to label AI-generated content, and how marketers can reduce their legal and reputational risks.

01 The quick answer

In most cases, businesses can continue using AI to create advertising content. Before publishing an AI-generated advert, ask four questions:

1. Does the content resemble a real person, place, object, organisation or event?
2. Could an ordinary viewer reasonably believe that the content is real?
3. Is a fictional person being presented as a genuine customer?
4. Has someone's face, voice, name or identity been used without clear permission?

If the answer to any of these questions is yes, the content may require a clear disclosure, additional consent or a legal review.

02 When do the new EU transparency rules take effect?

Article 50 of the EU Artificial Intelligence Act applies from 2 August 2026. It creates transparency duties for certain providers and deployers of AI systems. These include some interactive AI systems, generative AI tools, deepfakes and AI-generated text about matters of public interest.

The purpose of the rules is to help people recognise when they are interacting with an AI system or viewing certain types of artificially generated or manipulated content.

A limited grace period applies to some generative AI systems that were placed on the market before 2 August 2026. For those systems, the machine-readable marking requirement under Article 50(2) applies from 2 December 2026. This does not delay all of the other transparency duties.

Content generated before 2 August 2026 does not have to be labelled retrospectively under Article 50, although the European Commission encourages voluntary labelling where reasonably possible.

03 Have AI-generated UGC adverts been made illegal?

No. The EU AI Act does not introduce a general ban on using AI in advertising. Brands may still use AI to create:

• advertising copy;
• product images;
• video backgrounds;
• virtual presenters;
• synthetic voiceovers;
• product demonstrations;
• social media videos;
• animations; and
• other creative content.

The main issues are transparency, consent and whether the advert could mislead the audience. The legal risk becomes greater when an advert is designed to make synthetic content look like a genuine customer experience, an endorsement from a real person, an authentic event, an independent product review or an unedited recording of reality.

04 What is an AI-generated UGC advert?

UGC means user-generated content. Traditional UGC advertising usually shows a customer, creator or influencer speaking naturally about a product. It is often filmed in a personal style, using a phone camera rather than a large professional production.

An AI-generated UGC advert copies this style but uses artificial intelligence to create or change some or all of the content. Examples include:

• an AI avatar presenting a skincare product;
• a fictional person explaining the benefits of an app;
• an AI-generated voice placed over product footage;
• a presenter's face being altered with AI;
• a virtual influencer promoting a fashion product; or
• a testimonial-style video in which no real customer took part.

The technology itself is not necessarily the problem. The risk arises when viewers are encouraged to believe that a synthetic customer, experience or endorsement is genuine.

05 The AI tool and the brand may have different responsibilities

One of the most important parts of Article 50 is the distinction between a provider and a deployer.

What is a provider?

A provider is normally the person or company that develops an AI system, or has it developed, and then places it on the market or puts it into service under its own name or trade mark. For example, the company offering an AI video-generation platform may be the provider of that system.

Providers of systems that generate synthetic text, images, audio or video must generally ensure that their outputs are marked in a machine-readable format and can be detected as artificially generated or manipulated. This could involve metadata, digital watermarks, content credentials or other technical detection methods. The technical solution must be effective, interoperable, robust and reliable as far as technically feasible.

There is an exception where the system only performs an assistive function for standard editing or does not substantially change the input or its meaning.

What is a deployer?

A deployer is a person or organisation using an AI system under its authority for professional purposes. Depending on the circumstances, this could include:

• a brand publishing an advert;
• a marketing agency;
• a Shopify or e-commerce store;
• a content production company;
• an in-house social media team; or
• the business that commissioned the campaign.

In situations covered by Article 50, deployers may need to give the audience a clear and understandable disclosure. A hidden watermark or metadata field does not always satisfy this responsibility. For deepfakes and certain AI-generated text, the disclosure must be clear and perceivable to the person viewing or reading the content.

06 When must the use of AI be disclosed?

Not every use of AI requires a large visible label. The answer depends on the type of content, how realistic it appears, what it represents, how it is presented and what an ordinary viewer is likely to believe.

1. When a person is directly interacting with an AI system

Providers of AI systems intended to interact directly with people must design them so that users are informed they are interacting with AI, unless this is obvious from the circumstances. This can apply to customer service chatbots, interactive AI assistants, conversational shopping agents and interactive virtual presenters.

Simple disclosure: "You are speaking to our AI assistant."

This duty concerns genuine direct interaction between a person and an AI system. It does not necessarily apply to every system operating in the background of a website or business.

2. When the content is a deepfake

Under the EU AI Act, a deepfake is AI-generated or manipulated image, audio or video content that resembles existing people, objects, places, organisations or events and would falsely appear to a person to be authentic or truthful.

A video may therefore be a deepfake when it closely resembles a real person, place or event, shows something that did not actually happen and could reasonably be mistaken for authentic content. For example, a video showing a real person saying words they never said is likely to fall within this definition.

Deployers using AI to generate or manipulate content that constitutes a deepfake must disclose that the content has been artificially generated or manipulated.

3. When a fictional character is presented as a real customer

Imagine that an AI avatar says: "I used this serum for three weeks and my skin completely cleared up." If the person does not exist and has never used the product, this may create a fabricated customer experience.

Adding a label saying "AI-generated character" does not necessarily solve the problem. The statement still suggests that a real product experience took place.

EU consumer law prohibits misleading commercial practices and specifically addresses false consumer reviews and endorsements. Directive (EU) 2019/2161 prohibits submitting or commissioning false consumer reviews or endorsements, as well as misrepresenting consumer reviews or social endorsements to promote products.

Safer wording: "This serum contains vitamin C and is designed for daily use, according to the manufacturer's instructions."

This is very different from saying, "I used it myself and achieved these results." The first is a brand message. The second is a claimed personal experience.

4. When AI-generated text covers a matter of public interest

Deployers must clearly label certain AI-generated or manipulated text published for the purpose of informing the public about matters of public interest. This requirement may apply where the text is published, intended to inform the public, about a matter of public interest and not subject to meaningful human review or editorial control.

Matters of public interest may include politics, public health, public safety, fundamental rights, the environment, the economy, finance, science and other issues capable of affecting the public.

The requirement does not generally apply where the content has undergone proper human review or editorial control and a person or organisation accepts editorial responsibility for it. A simple spelling or grammar check is not enough. The Commission's guidance says that superficial or purely formal checks are not considered meaningful human review or editorial control. A reviewer should genuinely examine the accuracy of the content, its claims, the evidence, the sources and whether it is suitable for publication.

5. When someone's real face or voice is used

If an AI system is used to copy or recreate the face or voice of a founder, influencer, actor, employee or customer, the organisation should clearly document the person's permission. Important questions include:

• Did the person agree to the creation of an AI version of their face or voice?
• Does the permission cover paid advertising?
• Does it cover use across the EU?
• How long does the permission last?
• Can the brand generate new words or actions that the person never recorded?
• Can the material be reused in future campaigns?
• Can third-party agencies or platforms access the synthetic model?

An AI disclosure does not itself give a business permission to use someone's identity. Consent, contractual rights, data protection, intellectual property and national laws may all need to be considered separately.

07 Where should an AI disclosure appear?

Where Article 50 requires a visible disclosure, it should be clear, distinguishable and accessible. The European Commission's guidance on the optional EU AI icons recommends that the disclosure should be perceivable no later than the first time a person is exposed to the relevant content.

As good practice, the disclosure should:

• be visible against the background;
• use plain language;
• avoid technical jargon;
• appear at the beginning or first point of exposure;
• remain visible long enough to be read;
• work when the content is viewed without sound;
• not be hidden in terms and conditions; and
• remain available if the content is downloaded or reshared.

For video content, it may be sensible to use both an on-screen label and an audio disclosure. A disclosure hidden only in metadata may help with technical detection but will not necessarily meet a deployer's duty to inform the audience.

08 Example AI disclosure wording

The correct wording depends on the content and context. The examples below are practical suggestions, not mandatory legal wording.

The European Union has also developed optional icons for identifying AI-generated and AI-modified content. The icons are optional, but the legal labelling requirements under Article 50 are not. Using an EU icon does not automatically prove that the content complies with the law. The related Code of Practice is a voluntary compliance tool rather than a replacement for Article 50.

09 Examples of good and bad practice

A fabricated customer

An unauthorised influencer endorsement

Adding the words "AI-generated" does not resolve the central issue where a real person's identity and reputation are being used to create an endorsement that never happened.

An unrealistic product result

A disclaimer should not be used to hide or correct a fundamentally false claim. The underlying presentation must still be fair and accurate.

10 Can brands use Higgsfield, Veo, Runway, Kling or Synthesia?

Yes. The legal position is not determined by the name of the tool. No AI tool is automatically "legal" or "illegal" for advertising. The important questions are what the tool generated, whether it resembles reality, whether the audience could be misled, whether real identities were used, whether product claims are accurate, whether the correct disclosure was provided and whether the necessary commercial rights exist.

Before using any service commercially, review its current terms and conditions. Check commercial-use rights, ownership or licensing of outputs, restrictions on advertising, rules on face and voice cloning, responsibility for third-party rights and whether metadata or content credentials remain attached after export. These terms can change, so they should be checked for each campaign rather than assumed from previous use.

11 Does the EU AI Act apply to UK brands and agencies?

Potentially, yes. The United Kingdom is no longer an EU Member State, but that does not mean every UK business falls outside the EU AI Act. The Act can apply to providers and deployers established outside the EU where the output produced by an AI system is used within the European Union.

A London-based agency may therefore need to consider the EU AI Act where:

• it creates advertising for an EU-based client;
• a campaign is shown to consumers in EU countries;
• the AI output is used within the EU; or
• the service is specifically aimed at the European market.

UK businesses may also need to consider separate UK rules relating to advertising, consumer protection, data protection, intellectual property and the use of personal identities. Compliance with the EU AI Act does not automatically establish compliance with UK law, and the reverse is also true.

12 Must every AI image or video be visibly labelled?

Not necessarily. There are two separate concepts to understand:

1. Machine-readable marking by providers
2. Visible disclosure by deployers

Providers generally have to enable the detection of synthetic content generated or manipulated by their systems. However, the visible disclosure duty for deployers under Article 50(4) is more specific. It focuses mainly on deepfakes and certain AI-generated or manipulated text about matters of public interest.

Standard editing functions may also fall outside the provider marking requirement where they do not substantially change the input or its meaning. Examples may include basic colour correction, noise removal, simple image enhancement, minor lighting adjustments, standard audio cleaning or other assistive editing that does not materially change the content.

Content that is obviously artistic, fictional, creative or satirical may also be subject to more proportionate disclosure requirements, particularly where a stronger label would interfere with the enjoyment of the work. However, describing an advert as "creative" does not automatically exempt it. The context and likely understanding of the audience remain important.

13 Does advertising copy written with ChatGPT need an AI label?

Article 50 does not create a general visible-labelling requirement for every piece of commercial copy written with an AI tool. The special text-labelling duty relates to AI-generated or manipulated text that is published to inform the public about matters of public interest and has not undergone proper human review or editorial control.

Ordinary product copy, headlines and social captions will not necessarily fall within that category. However, the brand remains responsible for factual accuracy, product claims, advertising compliance, avoiding fabricated reviews, avoiding unsupported medical or financial claims, checking intellectual property issues and approving the final content. Using AI does not transfer responsibility for the advert to the software provider.

14 Can you clone your own voice?

Using an AI-generated version of your own voice is generally less risky than copying another person's voice without permission, but several questions should still be considered:

• Could the audience be misled about whether the recording is genuine?
• Does the content include other people's personal information?
• Does the publishing platform require disclosure?
• Were any third-party recordings used to create the voice model?
• Could the synthetic voice later be used outside the agreed campaign?
• Who can access, export or reuse the voice model?

Clear disclosure: "This voice was generated using an AI model of the speaker's voice."

15 Can brands use AI influencers?

Yes. Virtual and AI influencers are not generally prohibited. A brand should nevertheless make sure that:

• the character is not falsely presented as a real person;
• paid advertising is clearly identified;
• the character does not invent a personal product experience;
• it does not misleadingly resemble a particular real person;
• product claims are supported by evidence; and
• its virtual nature is clear to the audience.

Suitable account description: "Virtual AI character created by [Brand Name]."

The advertising disclosure and the AI disclosure serve different purposes. "Ad" tells the audience that content is commercial, while "AI-generated character" tells the audience that the person or content is artificial. One label does not necessarily replace the other.

16 AI advertising checklist for brands and online stores

Before publishing an AI-generated advert, review the following points.

Content

• Is the person shown real, synthetic or partly modified?
• Does the advert look like a genuine customer testimonial?
• Does it claim that someone personally used the product?
• Could an ordinary viewer mistake the content for reality?

Consent and identity rights

• Do you have written permission to use the person's face or voice?
• Does the permission cover paid advertising?
• Does it cover all intended countries?
• Is the length of the licence clear?
• Does the person understand that AI may create new words or actions?
• Can agencies, contractors or platforms reuse the synthetic material?

Disclosure

• Is the disclosure visible at the first point of exposure?
• Is it written in plain language?
• Can it be understood without sound?
• Does it remain visible for long enough?
• Is it more than a hidden metadata field or buried disclaimer?
• Will it remain visible if the content is reshared?

Advertising claims

• Is every factual claim accurate?
• Can results-based claims be proved?
• Is the advert creating a false customer experience?
• Are before-and-after visuals representative?
• Are health, financial or environmental claims properly supported?

Platform requirements

• Have you reviewed the platform's current AI-content rules?
• Does the platform offer its own AI disclosure setting?
• Is a separate "Ad" or "Sponsored" label required?
• Will uploading or compressing the file remove its metadata?

Record keeping

• Record the AI tool used.
• Keep the original and final files.
• Keep prompts and production notes where appropriate.
• Store signed permissions and disclosure wording.
• Keep evidence supporting advertising claims.
• Record who reviewed the content and when it was approved.

17 What are the possible penalties?

Article 50 is mainly enforced by national market-surveillance authorities. The EU AI Office and the European Data Protection Supervisor may also have roles in particular circumstances.

Certain breaches of the AI Act can result in administrative fines of up to 15 million euros or, for a business, up to 3 per cent of its total worldwide annual turnover for the previous financial year, whichever is higher. The rules also require proportionality to be considered for small and medium-sized businesses and small mid-cap companies.

Regulatory fines are not the only risk. A misleading AI advert could also result in removal of the campaign, rejection of paid advertisements, suspension of an advertising account, consumer complaints, contractual disputes, reputational damage and loss of customer trust.

18 Conclusion

The EU AI Act does not ban the use of artificial intelligence in advertising. Brands may continue using AI to create video, avatars, voiceovers, product images, advertising copy and social media content. From 2 August 2026, however, certain uses of AI are subject to clearer transparency requirements.

The more closely synthetic content resembles reality, and the more likely it is to mislead an ordinary viewer, the more carefully it should be reviewed and disclosed.

The safest approach is to use AI as a production tool, not as a way to manufacture false trust. Clear disclosure can reduce legal risk, but it can also help audiences feel more confident about the brands using this technology.

19 Frequently asked questions

20 Footnotes and references

1. European Parliament and Council of the European Union. Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence. Official Journal of the European Union, 13 June 2024. eur-lex.europa.eu
2. European Commission. Guidelines on transparency obligations for providers and deployers of AI systems. Published 20 July 2026; last updated 29 July 2026. digital-strategy.ec.europa.eu
3. European Commission. Transparency obligations under Article 50 of the AI Act: Questions and Answers. Last updated 24 July 2026. digital-strategy.ec.europa.eu
4. European Commission AI Act Service Desk. Article 50: Transparency obligations for providers and deployers of certain AI systems. ai-act-service-desk.ec.europa.eu
5. European Commission AI Act Service Desk. Article 3: Definitions, including the definitions of "provider", "deployer" and "deep fake". ai-act-service-desk.ec.europa.eu
6. European Commission. Unfair Commercial Practices Directive. commission.europa.eu
7. European Parliament and Council of the European Union. Directive (EU) 2019/2161, particularly Annex I, points 23b and 23c concerning consumer reviews and endorsements. eur-lex.europa.eu
8. European Commission. EU Icons for labelling AI-generated content. Last updated 20 July 2026. digital-strategy.ec.europa.eu
9. European Commission AI Act Service Desk. Article 2: Scope. ai-act-service-desk.ec.europa.eu
10. European Commission AI Act Service Desk. Article 99: Penalties. ai-act-service-desk.ec.europa.eu
11. European Commission. Quick Facts: Transparency rules for AI systems. Published 20 July 2026. digital-strategy.ec.europa.eu
12. European Commission. Code of Practice on Transparency of AI-Generated Content. Published 20 July 2026. The Code is a voluntary compliance tool, while Article 50 itself creates legal obligations. digital-strategy.ec.europa.eu