03.07.2026
Gernot Fritz, Tanja Pfleger, Fabian Duschnig
On 29 June 2026, the Council of the European Union gave the green light to simplify the AI rules under the Digital Omnibus on AI, also referred to as the Omnibus VII package (press release). This sets the legislative direction: the key application dates for high-risk AI systems will be postponed, certain obligations will be clarified, and the interplay between the AI Act and sector-specific product legislation will be recalibrated.
Background: Why an Omnibus?
The AI Act originally provided that key obligations for many high-risk AI systems would apply from 2 August 2026. In practice, however, it quickly became apparent that this timeline would be difficult to meet. Important European standards, technical specifications and guidance were either not yet available or only partially available.
For businesses, this created uncertainty as to how the new AI requirements should be aligned with existing sector-specific regulatory frameworks, for example for medical devices, toys or lifts. This is precisely where the Omnibus VII package comes in: it aims to reduce duplication, adjust timelines and facilitate the implementation of the AI Act.
New application dates for high-risk AI systems
The most significant practical change concerns the application dates for high-risk AI systems. The previous deadline of 2 August 2026 will no longer apply to stand-alone high-risk AI systems under Annex III of the AI Act. This category includes, for example, AI systems used in employment, education, creditworthiness assessments, law enforcement, critical infrastructure, or access to and enjoyment of essential private and public services.
Two new deadlines will apply instead: for stand-alone high-risk AI systems under Annex III, the relevant date will be 2 December 2027. For high-risk AI systems embedded in products that fall under certain sector-specific EU product safety legislation listed in Annex I, the relevant date will be 2 August 2028.
The basic regulatory model for high-risk AI systems remains unchanged. The core requirements — including risk management, data and documentation requirements, human oversight, accuracy, robustness, cybersecurity and quality management — remain in place. At the same time, the Omnibus package introduces targeted clarifications and simplifications, in particular regarding the interaction with sector-specific product legislation, technical documentation, quality management and conformity assessment.
New prohibited AI practices
The Omnibus package also expands the list of prohibited AI practices. Newly covered are, in particular, AI systems that generate or manipulate realistic intimate or sexually explicit content of an identifiable person without that person’s explicit consent. This includes, for example, deepfakes depicting intimate body parts or digitally removing clothing. AI systems used to generate or manipulate depictions of child sexual abuse will also be prohibited.
The new prohibitions will apply from 2 December 2026. The existing prohibitions under Article 5 of the AI Act have applied since 2 February 2025 and remain unchanged.
Shorter transition period for certain transparency obligations
There will also be an important adjustment to the transparency obligations. For AI systems, including general-purpose AI systems, that were placed on the market before 2 August 2026 and generate synthetic audio, image, video or text content, the new deadline for implementing Article 50(2) of the AI Act will be 2 December 2026.
The obligation to technically mark artificially generated or manipulated content therefore remains a key issue. Providers of such systems should assess whether their existing systems already include suitable technical solutions or whether they will need to be upgraded by December 2026.
Clarification of supervisory responsibilities
The Omnibus package also clarifies the allocation of responsibilities between the European Union’s AI Office and national supervisory authorities.
This is particularly relevant for AI systems based on general-purpose AI models where the model and the system are developed by the same provider. At the same time, the package expressly identifies areas in which national authorities will remain responsible, such as law enforcement, border management, the administration of justice and financial institutions.
Improved alignment with sector-specific product legislation
Another objective of the Omnibus VII package is to avoid duplicative regulatory structures. This is particularly relevant in areas already covered by sector-specific EU product legislation, such as medical devices, toys, lifts or watercraft.
Where such sector-specific rules contain comparable AI-related requirements, the application of certain AI Act requirements may in future be restricted by implementing acts. The aim is to avoid parallel or conflicting obligations.
For manufacturers and providers of regulated products, this is an important step. The high-risk obligations will not disappear. However, they should become better aligned with existing conformity assessment, safety and documentation requirements.
AI literacy
The AI literacy obligation under Article 4 of the AI Act has applied since 2 February 2025. The Omnibus package, however, adjusts the substance of this obligation: providers and deployers will in future be required to take appropriate measures to promote AI literacy. The focus therefore shifts more clearly towards appropriate organisational measures and less towards ensuring a specific level of competence.
Consequences for businesses
The Omnibus VII package provides greater clarity for the implementation of the AI Act. However, the additional transition periods should not be seen as a reason to further postpone compliance measures. Classifying AI systems, setting up risk management processes, preparing technical documentation, carrying out the necessary conformity assessments and establishing robust AI governance all require significant lead time.
No further postponements have currently been adopted. Businesses that prepare in a structured manner now will reduce implementation pressure and create a solid foundation for their AI compliance.
We are happy to support you in classifying your AI systems, developing a robust compliance roadmap and translating it into practical business processes.
