The European Union’s AI Act sets graduated obligations according to the risk of AI systems. Most ordinary advertising is not high-risk, but there are transparency obligations for certain uses, such as chatbots and synthetic content that may be mistaken for real, and prohibited practices the advertiser must know.
The essentials in a definition
The AI Act and advertising correspond to the obligations the AI Act creates for those who develop or use AI systems in marketing, including transparency and prohibited practices, in connection with European AI policies and the Unfair Commercial Practices Directive.
The risk-tier logic
The AI Act distinguishes prohibited practices, high-risk systems, transparency obligations for certain systems and minimal-risk uses. Obligations apply gradually, with different dates. Most AI uses in marketing (text generation, campaign optimisation) fall into limited or minimal risk, but it is worth confirming the framing of each case and application dates in official sources, including European AI policies.
This content is informational and does not replace legal advice.
Transparency for advertisers
When people interact with an AI system (for example, a customer-service assistant), they must be told it is AI, unless obvious. Synthetic content (image, audio, video) that realistically depicts people, places or events must be identified as generated or manipulated, with exceptions and conditions that vary. System providers and professional users have different obligations.
The Unfair Commercial Practices Directive continues to prohibit misleading consumers, with or without AI.
From AI inventory to compliance policy
- Inventory AI uses in marketing
- Classify risk and required transparency
- Define policies and owners
- Follow dates and guidance
Practices to avoid
The regulation prohibits certain practices, such as subliminal manipulation techniques or those exploiting vulnerabilities in ways that cause significant harm. In advertising, avoid targeting or messages that exploit people’s vulnerabilities (by age, economic situation or disability) harmfully. Assess the AI tools you use and ask providers for compliance information.
Review providers’ contracts and documentation.
Whoever uses AI in advertising should know what they use, for what and with which obligations.
Prepare the organisation
Inventory AI uses in marketing and advertising, classify risk, assign owners and internal policies (human review, disclosure, records). Train teams. Follow application dates and guidance from the Commission and national authorities, and update the policy accordingly.
Document decisions to be able to demonstrate diligence.
Decision table
| Use | Likely tier | Main obligation |
|---|---|---|
| Text generation and variations | Limited or minimal risk | Review and truthfulness |
| Customer-service chatbot | Transparency | Inform that it is AI |
| Realistic synthetic image or video | Transparency | Identify as generated or manipulated |
| Manipulation or exploiting vulnerabilities | Prohibited | Do not use |
Common mistakes to avoid
- Assuming it does not apply. Transparency obligations may exist.
- Relying only on the provider. The professional user has duties too.
- Not documenting. Diligence is hard to demonstrate.
How to measure the result
Track the updated inventory, approved policies and compliance with disclosure rules in samples.
Limits and exceptions
The regulation applies in phases, and guidance and interpretations keep emerging. This content is informational and not legal advice. When in doubt, consult a professional and official sources.
Illustrative example
Illustrative example: an agency inventories the AI tools it uses (text generation, image, chatbot), classifies risk, defines that the chatbot identifies itself as AI and that realistic synthetic images are labelled. It asks providers for compliance information and reviews the policy every quarter.
Frequently asked questions
Does the regulation already apply?
It applies in phases; confirm dates and obligations in official sources.
Are my ads high-risk?
Generally not, but assess the case and providers.
Do I need to label AI-generated images?
It depends on content type and conditions; when in doubt, be transparent.
Who should be responsible?
Someone with authority, supported by compliance and legal.
How do I follow changes?
By following official sources and reassessing the policy periodically.
Inventory AI uses in marketing, classify risk and transparency, define policies and owners, and follow the AI Act’s dates and guidance.
Sources and recommended reading
- Artificial Intelligence Act (AI Act)
- European Commission: regulatory framework for AI
- Unfair Commercial Practices Directive (2005/29/EC)
Last editorial review: October 2026. Informational content; always confirm current conditions in the official sources before deciding.