To overcome this regulatory deadlock, between May 6 and 7, 2026, the European Parliament and the Council of the EU reached a provisional political agreement on the Digital Omnibus AI legislative package.
This intervention does not overturn the sanctioning framework or the original principles of Regulation (EU) 2024/1689, but rather introduces a necessary technical and temporal recalibration. The European legislator's objective is to prevent the operational standstill of enterprises in the face of imminent deadlines and to streamline procedural requirements where bureaucratic duplications arise. From a legislative process perspective, the recently reached political agreement must be formally adopted and published in the Official Journal by August 2, 2026.
For internal legal departments and corporate General Counsels, a strategic window of opportunity opens to reshape the compliance roadmap. The main new profiles are analyzed below.
Recalibration of the risk timeline (HRAIS)
The update with the greatest impact lies in the postponement of deadlines for high-risk artificial intelligence systems (HRAIS). Overcoming the previous and complex conditional mechanism, the agreement sets the following terms:
- "Stand-alone" Systems (Annex III): compliance obligations for software applied to sensitive sectors, including human resources, credit scoring, biometrics, and education, are deferred to December 2, 2027.
- Systems Integrated into Regulated Products (Annex I): for AI systems embedded in medical devices, industrial machinery, or the automotive sector, the final deadline for alignment is postponed to August 2, 2028.
This temporal extension responds to the need, expressed by both national authorities and industry, to await the issuance of official guidelines from the European AI Office and the publication of harmonized technical standards, which are indispensable for the practical application of compliance requirements.
Simplification and rationalization of compliance
On a procedural level, the Digital Omnibus AI intervenes to eliminate the risk of so-called "double compliance". If an AI system is integrated into an asset already subject to strict European safety regulations (such as the Machinery Regulation), the conformity assessment procedures will be centralized.
Furthermore, there is a narrowing of the notion of "Safety Component": software used exclusively for assistive functions or mere internal quality control will not automatically fall under the high-risk classification, thereby reducing the administrative burden on manufacturers. Finally, the flexibility and exemptions initially reserved solely for SMEs are extended to the category of small mid-caps.
New prohibitions and short-term requirements
Faced with a relaxation on high-risk systems, the Omnibus agreement tightens security safeguards on other fronts, introducing a restrictive amendment to Article 5 of the AI Act. A strict prohibition is established on the development and commercialization of generative AI systems aimed at creating non-consensual intimate content ("Nudifiers") or child sexual abuse material (CSAM). Under civil and criminal liability profiles, the burden of proof and the obligation to implement preventive technical filters will fall directly on the model provider, provided that such illicit use is "reasonably foreseeable".
Regarding transparency (Article 50), the date of August 2, 2026, remains confirmed for general information obligations toward users, while a period of time until December 2, 2026, is granted for the integration of digital markers (watermarking) on content generated by systems already placed on the market.
Operational conclusions for management
The postponement of deadlines must be interpreted by enterprises as an opportunity to structure a solid and integrated AI Governance model. The immediate priority for corporate boards consists in mapping the systems in use, verifying the resilience of transparency requirements in view of the August 2026 deadline, and planning the alignment with the new safeguards of 2027 and 2028.
van Berings is at the disposal of enterprises to support them through the complex transition toward the new regulatory framework, offering strategic counsel aimed at structuring effective, secure governance models and compliance programs aligned with European deadlines.
