Transparency Requirements for AI Systems
On July 20, 2026, the European Commission published guidelines on the implementation of transparency obligations for certain AI systems under Art. 50 AI Act. Providers and deployers of certain AI systems must, in specific cases, transparently inform users about the use of AI. This may be required, for example, where an AI system interacts directly with people, where images and videos are generated using AI (deepfakes), or where certain AI-generated text publications of public interest are involved. The newly published draft is intended to help organisations better comply with the transparency obligations under the AI Act in practice. The draft guidelines were designed in parallel with the Code of Practice for the labelling of AI-generated content, published on June 10, 2026.
The voluntary Code of Practice, which is currently being assessed by the European Commission and the AI Board for adequacy, is intended to help organisations implement these transparency obligations under the AI Act — particularly regarding the labelling and recognisability of AI-generated or manipulated content. The Code is aimed both at providers of generative AI systems and at deployers using such systems. It contains practical rules on machine-readable markings, technical recognisability and clear notices to users.
Participation in the Code remains voluntary, but may offer advantages for organisations: by implementing the measures set out in the Code, they can demonstrate compliance with the transparency requirements of the AI Act. This creates greater legal certainty and predictability. It also reduces administrative effort, as the adequacy of each alternative measure does not have to be demonstrated individually. Organisations can sign the Code of Practice here and thereby confirm that they comply with the standards set out therein. In principle, providers and deployers can sign the Code at any time (FAQs). After August 2, 2026, organisations that want to sign the Code will have to demonstrate compliance with these requirements of the AI Act. Your team of consultants will be happy to support you in assessing whether participation in the Code of Practice makes sense for you.
On May 19, 2026, the European Commission also published draft guidelines on the classification of high-risk AI systems. The draft is intended to support providers and deployers of AI systems as well as market surveillance authorities in assessing whether an AI system qualifies as a high-risk AI system under the AI Act and thereby ensure consistent application of the requirements. The draft explains how the Commission interprets certain concepts for classifying AI and contains practical examples that may help with classification. The examples are intended to cover as many sectors and use cases as possible; however, they are not exhaustive and may be amended in the future.
Unlike the Code of Practice, both guidelines are currently only drafts and still need to be finalised. Nevertheless, both drafts are highly relevant for the practical implementation of the AI Act, as they specify legal requirements in more detail. We recommend reviewing existing and planned AI systems with regard to transparency obligations and classification as high-risk AI against the background of these documents. We will continue to monitor the finalisation of the guidelines and inform you about any specific need for action.
We are pleased to assist you with the legal analysis and practical implementation. You can contact us directly through our Contact page.
