What has the Digital Omnibus on AI changed in practice?

Publications

FILTRER

24 août 2026, VdA

Artificial Intelligence Act

The AI Omnibus introduced changes that may affect how organisations assess, govern and use AI systems.

Updated application dates: The application of several AI Act obligations has been postponed. The rules applicable to stand-alone high-risk AI systems will now apply from 2 December 2027, while the rules for high-risk AI systems embedded in products or constituting safety components of products will apply from 2 August 2028. The deadline for Member States to establish AI Regulatory Sandboxes has also been postponed to 2 August 2027.
Transparency obligations: The application of the providers' marking and detection obligations relating to AI-generated content and deepfakes has been postponed until 2 December 2026. However, the transparency obligations applicable to deployers of AI-generated content, deepfakes, emotion recognition and biometric categorisation systems, as well as the obligations applicable to providers of AI systems that interact directly with natural persons, continue to apply from 2 August 2026.
New prohibited AI practice: The AI Omnibus expressly prohibits AI practices involving the generation of non-consensual intimate content and AI-generated child sexual abuse material (CSAM).
Bias detection: The requirement of strict necessity for the processing of special categories of personal data for the purposes of detecting and correcting bias in AI systems has been reinstated.
Sector-specific legislation: Explicit reference to guidance on the interaction between the AI Act and harmonised sectoral legislation, including legislation on medical devices, machinery and toys, helping to avoid duplicative compliance requirements.
AI Regulatory Sandboxes: The deadline for Member States to establish AI Regulatory Sandboxes has been extended until 2 August 2027, providing additional time to develop controlled testing environments for innovative AI systems.
AI Office competences: The supervisory role of the AI Office over AI systems based on general-purpose AI models has been clarified, while confirming the areas that remain within the competence of national authorities.
European Commission guidance: The European Commission is required to publish guidance to support providers and deployers of high-risk AI systems subject to harmonised sectoral legislation, promoting a more consistent and practical application of the AI Act across sectors.