Guidelines on Transparency Obligations Under the EU AI Act
Summary
On 20 July 2026, the European Commission published the Guidelines on transparency obligations for providers and deployers of certain AI systems under the EU AI Act. Those obligations come into force on 2 August 2026, except for AI systems (including general-purpose AI systems) generating synthetic audio, image, video or text content—these benefit from a short extension to 2 December 2026.
The Guidelines clarify the scope of the obligations under Article 50 of the AI Act, providing details on how the rules may be interpreted as well as practical examples of what providers and deployers may encounter in their efforts to comply with the obligations. The Guidelines are non-binding (as any binding interpretation of the AI Act may be given only by the Court of Justice of the European Union), but they set out certain useful context and recommendations.
The purpose behind the AI Act’s transparency obligations was to minimise the risks of misinformation, fraud, manipulation at scale, impersonation and deception. The obligations aim to foster trust in AI tools and safeguard the integrity of the information ecosystem, allowing individuals to be aware when they are dealing with AI and to calibrate their reliance accordingly. The obligations on providers include ensuring transparency regarding AI systems intended to interact directly with individuals and AI systems generating synthetic audio, image, video or text content. The obligations on deployers of AI tools include emotion recognition systems or biometric categorization systems and AI systems that generate or manipulate image, audio or video content constituting deep fakes.
Compliance may be demonstrated by adhering to the recently adopted Code of Practice on Transparency of AI-generated content, or otherwise through adequate alternative means.
