In recent years, artificial intelligence has found ever wider application across many areas of business. This also applies to advertising agencies – AI tools now assist in generating graphics, music, and even entire advertisements and campaign scripts. This speeds up work and cuts costs, but it also raises several legal issues worth considering before a finished ad reaches the client and the market.
Who owns the copyright to an ad generated by an AI tool?
As a general rule, Polish copyright law protects only what results from human creative activity. Graphics or text generated entirely by AI, without a substantial human contribution, formally do not constitute a work at all. This means no one holds exclusive copyright to them. Most importantly, the entity that creates the ad will not hold rights to such material either. In practice this carries serious consequences – for example, a competitor could freely copy such a graphic and use it for another purpose without infringing any copyright.
What can be done to acquire copyright to AI-generated graphics? The automatic output of the AI tool should serve only as a starting point, not the finished product. What matters in this case is the selection and combination of elements and the user’s own edits – the more genuine creative human input goes into the final version of the graphic, the stronger the basis for calling it a work. And it is precisely a work that is protected by copyright.
Risk of infringing third-party rights through the use of AI in advertising
AI models are trained on enormous sets of other people’s images, texts, and photographs. As a result, the output may end up similar enough to an existing work, or to a specific person’s likeness (or even their voice), that it infringes third-party rights – personal rights, image rights, copyright, or industrial property rights (trademark protection).
For example, if AI generates a face resembling a well-known person, or a voice similar to a specific voice-over artist or actor, using it to create an ad without their consent will very likely infringe their rights. Liability in such a case falls (as a rule) on whoever published and commercially used the ad – most likely the advertising agency or the client, not the AI tool provider. That is why AI-generated material, especially anything resembling real people or other people’s work, should always be carefully checked.
New AI Act obligation – labelling AI-generated content
From 2 August 2026, the provisions of the EU Artificial Intelligence Act (Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024) concerning, among other things, the labelling of AI-generated content will become fully applicable. This mainly concerns material that could pass as authentic – realistic photos, videos, or voices resembling real people, objects, or events (so-called deepfakes). If an ad suggests we are seeing a genuine statement from an expert, client, or celebrity, when in fact it is AI-generated, this must be clearly and legibly indicated, at the latest when the recipient first encounters the material.
The consequences of ignoring this obligation can be severe. Fines can reach EUR 15 million or 3% of a company’s total global annual turnover – whichever amount is higher. This is no longer a matter of good practice, but a hard legal obligation.
Who is liable for an AI tool’s mistake?
AI can confidently state false information, invent a non-existent statistic, or formulate a promise that the client’s product does not fulfil. If such content ends up in an ad, liability – including for misleading consumers – rests, depending on the situation, with the agency or the client, not with the creator of the tool.
That is why, before AI-generated material is published, ordinary human verification of the claims it makes is needed. It is also worth agreeing with the client in advance (preferably in the contract) who is responsible for any errors, third-party rights infringements, or the cost of withdrawing a campaign. The absence of such arrangements does not eliminate the risk – it simply postpones the dispute until after the damage has already occurred.
The use of artificial intelligence in advertising is becoming common practice, but it is essential to know how to use it responsibly. The four pillars of safe use of AI tools are: conscious human input into the final project, verification that the result does not infringe third-party rights, awareness of the new AI content labelling obligations, and a clear division of liability between the ad’s creator and the client. Each of these elements can be settled in advance. That seems the sensible approach, given the cost of fixing a problem after the ad has already been published on the market.
