At a glance
- Art. 50 AI Act applies since 2 Aug 2026
- Label chatbots and deepfakes
- AI claims must be substantiated (AI washing)
Artificial intelligence is changing advertising at every level: images and videos are generated, product copy is automated, customer contact is handled by chatbots, prices are set dynamically. Legally, there is no special “AI advertising law”. The UWG, the AI Act, copyright, personality rights and data protection apply side by side.
Since 2 August 2026 the transparency obligations of Article 50 AI Act apply. For generative AI systems placed on the market before that date, a transition period until 2 December 2026 applies to machine-readable marking of outputs.
Transparency obligations under Art. 50 AI Act
- Chatbots and voice assistants (para. 1): users must be informed that they are interacting with AI, unless this is obvious. For sales and service chats, this means a clear notice at the start of the conversation.
- Marking of generated content (para. 2): providers of generative AI must mark outputs in a machine-readable way.
- Deepfakes (para. 4): deployers of AI that generates or manipulates realistic images, audio or video must disclose that the content is artificial. For evidently artistic, satirical or fictional works, a non-intrusive notice suffices. AI-generated text on matters of public interest must be disclosed unless it has undergone human review and editorial responsibility.
Breaches can attract fines of up to EUR 15 million or 3% of worldwide turnover. Because these obligations also protect consumers, there are good arguments that breaches are actionable by competitors as a breach of statutory duty under Section 3a UWG. Case law is still pending.
Prohibited practices (Art. 5 AI Act)
Since 2 February 2025, AI systems that deploy subliminal or purposefully manipulative techniques, or exploit vulnerabilities due to age, disability or social situation, to materially distort behaviour and cause significant harm are prohibited. Below that threshold, Section 4a UWG and the blacklist apply.
AI under unfair competition law
| Scenario | Risk | Provision |
|---|---|---|
| “AI-powered” without meaningful AI (AI washing) | misleading as to main characteristics | Section 5 UWG |
| realistic AI ad images (results, before/after) | misleading as to effects | Section 5 UWG, Section 11 HWG |
| synthetic testimonials | fake endorsements | Annex no. 23c UWG |
| AI-generated reviews | fake reviews | Annex nos. 23b, 23c UWG |
| virtual influencers without ad labelling | hidden commercial intent | Section 5a(4) UWG |
| chatbot errors on prices, warranty, delivery | misleading, attributed to the business | Sections 5, 8(2) UWG |
| personalised prices | information duty | Art. 246a § 1 no. 6 EGBGB |
| voice and image clones of celebrities | personality rights, exploitation of reputation | BGB, KUG, Section 4 no. 3 UWG |
In July 2026, the German Wettbewerbszentrale published updated guidance on AI-generated content, recommending, among other things, labelling realistic AI ad images in case of doubt, disclosing virtual influencers and clearly identifying chatbots at the outset.
Liability
A business is liable for advertising statements made by its AI systems as if an employee had made them. Under Section 8(2) UWG, the business owner is also liable for employees and agents. The Federal Court of Justice confirmed this agent liability in 2026 for automatically generated search ads where the business outsources ad placement and supplies the product data (BGH, judgment of 11 March 2026, I ZR 28/25).
Checklist
- Inventory of AI use in marketing, sales and service
- Labelling: chatbot notice, deepfake disclosure, internal rules for AI images
- Human review of all factual claims before publication
- Evidence: document what an “AI” claim actually refers to
- No synthetic customer voices or reviews
- Contracts with agencies and influencers: liability, rights, labelling
- Rights in training and output data, voices and faces
- Monitoring of chatbot outputs
