Transparency Obligation
The transparency obligation requires providers and deployers of AI systems to disclose that artificial intelligence is involved. It is set out in Article 50 of the EU AI Act and concerns AI systems of "limited risk" – above all chatbots and AI-generated content. Users should be able to recognise when they are interacting with an AI or looking at AI-generated content. The obligation becomes binding on 2 August 2026.
What the transparency obligation requires
Article 50 bundles several disclosure duties: deployers must design chatbots so that users recognise the AI before interacting (para. 1). Providers of generative AI must mark AI content in a machine-readable way (para. 2). And AI-generated or manipulated content – deepfakes in particular – must be disclosed as such (para. 4). The concrete implementation on the website is described under AI Labelling .
From when and for whom it applies
The transparency obligations under Article 50 become binding on 2 August 2026 – independently of the postponed high-risk regulation. There is no general exemption for small and medium-sized enterprises: anyone operating a chatbot or publishing AI content is affected. Within the EU AI Act's risk-based approach, the transparency obligation represents the "limited risk" class – stricter than the obligation-free minimal class, but considerably leaner than the high-risk requirements.