Liaise Advocaten logo

AI in advertising: what advertisers and agencies may do

A campaign image that comes out of an AI tool in a single afternoon. A voice-over without a voice actor, a model who never existed. Making advertising with AI is faster and cheaper, but the existing rules apply to it in full. Since 2 August 2026 the AI Act has added a disclosure duty for deepfakes. It reaches advertising more often than you might expect.

Liaise advises advertisers and agencies in the Netherlands that use AI in their campaigns. We review a campaign before it goes live and stand by you if a cease-and-desist letter or a complaint to the Dutch Advertising Code Committee arrives anyway.

Need legal advice straight away? Call 020 675 88 21.

Who owns the rights in an AI-generated campaign

Copyright only arises where a human makes free, creative choices that show in the result. With an image that an AI tool fills in by itself, that is a problem. On 13 February 2026 the Amtsgericht München refused protection for three logos made with AI, despite a prompt of 1,700 characters. That standard comes from the case law of the Court of Justice of the European Union and applies in the Netherlands just the same. Read how the Munich court ruled.

If a competitor copies your AI campaign image, copyright will often leave you with nothing to enforce. If you want a campaign you can defend, let a human make the decisive choices: a photographer who sets the composition, a designer who genuinely reworks the generated image. For a logo or a recurring campaign character, a trade mark registration is the stronger route. A trade mark does not have to be devised by a human.

More about AI and copyright →

When AI output infringes existing work

An AI tool can return a well-known character or an existing illustration almost verbatim. In February 2024 the Guangzhou Internet Court held the operator of a Chinese AI website liable because users could generate recognisable Ultraman figures with it. It was regarded as the first judgment worldwide on the output of an AI image generator. The operator had to pay 10,000 yuan and build in filters.

If, as an advertiser, you use an AI image containing recognisable parts of someone else’s work in your campaign, you infringe, because you are the one making it public. Not knowing where the tool got it from does not change the infringement, although it can affect the amount of damages. The Court of Justice drew the line more sharply on 4 December 2025 in Mio/konektra. There is infringement as soon as creative elements of a work have been recognisably reproduced, even if the two images leave a different overall impression.

So before publication, ask which tool an image was made with and whether the provider of that tool indemnifies business users against claims over the output. Some do, on conditions and usually only on a business plan.

A campaign in the style of a well-known illustrator

Style is free. As early as 1995 the Dutch Supreme Court held that merely following a fashion or style attracts no protection (Decaux/Mediamax). In Mio/konektra the Court of Justice confirmed that following a trend or artistic movement is not infringement, as long as no concrete creative elements are taken over.

With a prompt such as “in the style of”, the risk lies elsewhere. A prompt like that steers the tool towards the work of one artist, so the output often contains more than the style alone, such as a distinctive character or a composition you recognise from a well-known book. Those elements can be protected.

If you also name the artist in the advertisement, you quickly suggest that the artist worked on your campaign. If that is not true, it is misleading.

An AI lookalike of a Dutch celebrity is also a portrait

Max Verstappen took Picnic to court over a video featuring a lookalike. On 22 April 2022 the Dutch Supreme Court held that an image of a double can be a portrait of the person the double resembles, even if viewers realise it is not the real person (ECLI:NL:HR:2022:621). The same reasoning applies to a generated face, so you need consent if your AI model looks just like a well-known athlete or presenter. Expect to pay a fee that reflects the commercial value of that fame as well.

A voice does not fall under Dutch image rights. An imitated voice in a commercial can still be unlawful towards the person whose voice it is, especially if you link that person to your brand this way. How to cover image and voice in a campaign is set out on image rights for businesses and in our article on the use of voices by AI.

Which advertisements need an AI label since 2 August 2026

Article 50(4) of the AI Act requires whoever creates a deepfake to disclose that the image, video or audio is artificial, at the latest when someone first sees or hears it. A deepfake covers content that resembles existing persons, places or events and would falsely appear to be authentic.

For advertising, that is wider than it sounds. The guidelines the European Commission published on 20 July 2026 also count realistic AI personas. A generated model therefore does not have to resemble a real person to fall within the disclosure duty. The Commission itself lists as deepfakes a video in which an AI version of a well-known influencer promotes a product, and a realistic avatar of a CEO addressing staff.

The label is the job of the agency operating the AI tool. Under the guidelines, a company that merely commissions an agency is not a deployer, as long as it does not itself decide whether and how AI is used. If you, as the advertiser, require the campaign to be made with AI, that may be different.

For evidently creative or fictional work the disclosure may be lighter, for instance in the end credits. According to the Commission advertising can sometimes qualify, but not where the content is purely commercial and recognisable as such. For an ordinary advertisement, assume the label belongs visibly or audibly in the advertisement itself. A breach carries a fine of up to 15 million euros or 3 percent of worldwide annual turnover, the lower of the two for SMEs.

When you have to label AI content →

Where the AI label goes in every version of your campaign →

Misleading AI advertising under the Dutch Advertising Code

A label does not make an advertisement lawful by itself. Advertising may not mislead, with or without AI (article 8 of the Dutch Advertising Code and article 6:193c of the Dutch Civil Code). A generated “customer” enthusing about your product suggests an experience that never happened.

If you work with a virtual influencer, look at the Dutch Advertising Code for Social Media & Influencer Marketing. Since 1 July 2026 it states expressly that AI and virtual influencers may not mislead in advertising (article 4(e)). The company behind the account counts as the distributor of the advertising and must therefore also disclose its relationship with the advertiser.

Complaints about advertising go to the Dutch Advertising Code Committee. Anyone can file one, including a competitor.

What advertiser and agency agree on AI

Many of the risks above end up in the contract. That is where you settle whether the agency may use AI and with which tools, who adds the label and who bears a claim if a generated image turns out to resemble existing work. And what does the advertiser actually get if no copyright subsists in the result? If you agree nothing, a rights holder can go straight to the advertiser, because the advertiser is the one publishing the advertisement.

More about the AI clause →

AI in advertising lawyer

Liaise works for advertisers, agencies and creators in the Dutch advertising and media world. Before an AI campaign goes live, we look in the images for recognisable work of others and recognisable people. Then we look at the label and at the claims in the advertisement. We also put the agreements with your agency or your client on paper. If a cease-and-desist letter, a complaint to the Dutch Advertising Code Committee or interim injunction proceedings follow anyway, we act for you.

Working with AI in a campaign and want to know where you stand? Get in touch or call 020 675 88 21. Response within one working day.

Liaise Advocaten
Lawyer

Merel advises and litigates in the fields of media law, film law, copyright, privacy law, contract law and intellectual property law. She acts for established and emerging media personalities, producers in film, television and podcasts, victims of unlawful publications and creative entrepreneurs.

Liaise Advocaten
Lawyer

Charissa advises and litigates in the fields of media law, intellectual property, IT and AI law, privacy, e-commerce and corporate law. Press-related and journalistic matters, copyright, freedom of expression versus privacy and cross-border issues often play a central role in her work.

Frequently asked questions about AI in advertising

Unsure about AI in a campaign? We are happy to take a look.

  • Merel Teunissen
  • Charissa Koster
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

Response within one working day

Your message goes to one of our lawyers.

  • Merel Teunissen
  • Jaap Versteeg
  • Charissa Koster
  • Roland Wigman
  • Alexandra Iedema
  • David Allick

How can we help?

How do we reach you?