Digital Omnibus VII Package Delays AI Act Deadlines for High-Risk Systems

Companies preparing to roll out the EU's AI Act have been given more time to finalize procedures for high-risk artificial intelligence systems. The Digital Omnibus VII legislative package pushes back key deadlines by more than a year, though not every obligation has been deferred in the same way.

Digital Omnibus VII is the latest step in the EU's effort to simplify digital rules, including the AI Act itself. Rather than sticking rigidly to the original timeline, the European Commission has decided to split the rollout of high-risk system requirements into two stages, depending on whether a system operates standalone or is part of a larger physical product.

Two deadlines instead of one

For standalone high-risk AI systems, meaning those not integrated into any device or machine, the new implementation deadline is December 2, 2027. This covers systems used in biometrics, critical infrastructure, education, recruitment and employment, as well as migration, asylum and border-control procedures.

A different schedule applies to AI systems built into products regulated under separate EU rules, such as elevators or toys. For these, the deadline for meeting the requirements has been pushed to August 2, 2028. The distinction is practical, since physical products with embedded AI require longer certification and testing cycles than standalone software.

An exception to the delays

Not every change moves toward longer deadlines. The obligation to label content generated by artificial intelligence, one of the AI Act provisions most visible to ordinary internet users, has actually been moved up under the new schedule. It will take effect on December 2, 2026, sooner than originally planned for some of the other requirements.

That means companies publishing content created with generative AI, whether images, text or recordings, will need to implement labeling mechanisms for such material earlier than the requirements for high-risk systems themselves kick in. The regulator is effectively prioritizing transparency toward content audiences over the full rollout of technical requirements for system providers.

Why the regulator is granting more time

The authors of the Omnibus VII package argue that extending deadlines for high-risk systems gives companies a real chance to better prepare their processes, resources and internal allocation of responsibility. High-risk systems carry the heaviest regulatory burden in the entire AI Act, including technical documentation, risk-management systems, human oversight and registration in the EU database.

Critics of the original AI Act timeline have argued for months that many companies, especially smaller ones, would not be able to meet the requirements on time, partly due to the lack of finalized technical standards and implementing guidance from the European Commission. The postponement is therefore also an acknowledgment that progress on detailed guidelines has not kept pace with the ambitions of the original schedule.

What this means for Polish companies

For Polish businesses working on high-risk AI system deployments, the new schedule means extra time to prepare compliance documentation, though it is not a reason to halt work altogether. Companies that have already started AI Act audits should use the additional time to refine their processes rather than shelve them.

At the same time, the earlier deadline for labeling AI-generated content, effective as soon as December 2026, applies to a much broader group of businesses than just high-risk system providers. Companies publishing marketing materials or website content created with generative AI should begin implementing labeling mechanisms well in advance, regardless of whether the rest of the high-risk system rules apply to them.

Sources: Pakiet Omnibus VII - nowy harmonogram dla wdrożenia AI Act (kancelariamacura.pl), AI Act z nowym harmonogramem (rp.pl)