EU Digital Omnibus on AI Regulation enters into force, strengthening AI Office oversight
The European Union’s Digital Omnibus on AI Regulation has entered into force, clarifying the AI Office’s supervisory responsibilities, expanding its enforcement powers, and introducing new regulatory sandboxes to support AI innovation.
On 27 July 2026, the European Union’s Digital Omnibus on AI Regulation (2025/0359) entered into force, introducing amendments intended to simplify the implementation of the EU AI Act while clarifying governance arrangements for AI oversight across the Union.
A central element of the regulation is clarifying the AI Office’s supervisory role. Under the new rules, the AI Office has exclusive responsibility for overseeing AI systems built on general-purpose AI (GPAI) models when both the underlying model and the AI system are developed by the same provider or by providers belonging to the same corporate group. The regulation also extends the AI Office’s authority to AI systems embedded in, or constituting, very large online platforms (VLOPs) and very large online search engines (VLOSEs).
The regulation defines several exceptions to this supervisory framework. The AI Office’s exclusive competence does not apply to AI systems regulated under Annex I product legislation, certain high-risk AI systems listed in Annex III, AI systems used for law enforcement, border management, financial institutions under Article 74(6), or AI systems used in the administration of justice.
In addition to clarifying supervisory responsibilities, the Digital Omnibus strengthens the AI Office’s enforcement capabilities. The Office is granted new investigatory powers, including the ability to conduct on-site inspections, accept binding commitments from providers, and impose fines and periodic penalty payments in the event of non-compliance.
The regulation also introduces measures intended to support innovation and regulatory compliance. It authorises the AI Office to establish an EU-level regulatory sandbox for AI development, with priority access for small and medium-sized enterprises (SMEs), start-ups, and small mid-cap companies (SMCs). At the national level, member states are required to establish their own regulatory sandboxes by 2 August 2027, providing controlled environments where AI developers can test systems under regulatory supervision.
The Digital Omnibus forms part of the EU’s broader effort to streamline the implementation of its digital legislation while maintaining consistent oversight of high-risk and general-purpose AI systems.
