The AI Omnibus Regulation: What’s Next for the AI Regulatory Sandboxes?
By Pierre Berger, Joris Latui and Marieke van Gestel, DLA Piper
On 27 July 2026, the AI Omnibus Regulation took effect across the European Union. For AI regulatory sandboxes, the new rules bring several changes to their timing, organisation and governance. While the existing concept remains largely unchanged, the Regulation introduces several adjustments of which companies and authorities should be aware.
The European Commission presented the AI Omnibus Regulation1 as part of the Digital Omnibus package on 19 November 2025.
The package responds to practical challenges that emerged during the initial implementation of the AI Act2 and aims to make the framework clearer, more proportionate and supportive of innovation.
The AI Act entered into force on 1 August 2024 and follows a phased implementation. Against this background, the AI Omnibus Regulation introduces changes to testing and experimentation provisions, including AI regulatory sandboxes and real-world testing.
What is an AI regulatory sandbox?
The concept of an AI regulatory sandbox is not new. They have been used for years, particularly in the financial sector, to allow innovative companies to test new products and services in a controlled environment under regulatory supervision. The approach has since spread to other sectors.
The AI Act brings this concept explicitly into the AI regulatory framework. Article 3(55) defines an AI regulatory sandbox as: “a controlled framework established by a competent authority where providers or prospective providers can develop, train, validate and test innovative AI systems for a limited period under regulatory supervision, including, where appropriate, in real-world conditions.”
The objective is not to create a regulatory-free environment. Sandboxes allow companies to engage with regulators at an early stage, identify regulatory requirements and address potential risks before an AI system is placed on the market. This can provide greater regulatory clarity and help companies bring innovative products to market more efficiently.
The benefits are not limited to companies. Regulators also gain first-hand insight into emerging technologies and practical challenges faced by businesses. This can strengthen regulatory expertise, improve supervision and, over time, inform the development of future rules and guidance.
A new timeline for national sandboxes
The AI Omnibus Regulation does not fundamentally change the concept of regulatory sandboxes under Article 57 of the AI Act. One of the most practical changes concerns the deadline for the national AI regulatory sandbox.
Under the AI Act, Member States were expected to have at least one national AI regulatory sandbox operational by 2 August 2026. The AI Omnibus Regulation extends this deadline by one year, giving Member States until 2 August 2027. This additional time allows Member States to establish the necessary legal, institutional and supervisory structures.
A key requirement remains that the sandbox must be established by a competent national authority. That authority is responsible for overseeing the sandbox and providing guidance to participating companies on regulatory compliance during the testing phase.
An EU-level sandbox
National sandboxes are not the only development to watch. The Regulation also introduces an EU-level regulatory sandbox that may be established by the AI Office in cooperation with relevant authorities. Importantly, the establishment of an EU-level AI regulatory sandbox by the AI Office does not affect the Member States’ powers to establish and oversee regulatory sandboxes for AI systems under their supervision.3
This EU-level sandbox is intended to complement national initiatives rather than replace them. Its scope will be defined to avoid unnecessary overlap and to clarify the respective responsibilities of EU and national authorities. The EU-level framework could provide an additional channel for obtaining regulatory input at European level.
The Regulation places particular emphasis on access for smaller innovative businesses. SMEs, start-ups and small mid-cap companies will receive priority access to sandboxes established by the AI Office. This reflects the broader objective of ensuring that innovative businesses can access regulatory expertise when developing and testing new AI technologies.4
Broader cooperation and oversight
The AI Omnibus Regulation also broadens cooperation between authorities. Whereas the AI Act initially focused on cooperation between competent national authorities, the new Regulation explicitly involves the AI Office and the European Data Protection Supervisor.
Through the AI Board, these authorities can coordinate their work, share expertise and support the joint establishment and operation of regulatory sandboxes, including across different sectors. This broader cooperation and facilitates the exchange of best practices.5
Finally, the Commission’s power to adopt implementing acts on the establishment, development, implementation, operation and supervision of regulatory sandboxes will be extended to cover a new category, namely governance. These rules will set out detailed requirements for governance, including the involvement and oversight of data protection authorities as well as the coordination and cooperation between authorities at national and EU level.6
The Belgian perspective
Belgium currently remains in the early stages of establishing its national AI regulatory sandbox. At the time of writing, no Belgian AI regulatory sandbox is operational, and the competent authorities responsible for establishing and supervising such a framework have not yet been formally designated.
The idea of one or more Belgian AI regulatory sandboxes has nevertheless already been discussed in parliamentary sessions and working groups. In its General
Policy Note of 25 April 2025, the Minister for SMEs indicated that a legal study was underway to develop recommendations and draft legal instruments, with a view to ensuring compliance with the AI Act.7
Despite these efforts, no draft legislation providing the legal basis for a Belgian AI regulatory sandbox has been published to date.
The extension of the EU deadline to August 2027 therefore gives Belgium additional time to put the necessary framework in place.
Looking ahead
For companies, particularly start-ups and SMEs, sandboxes could become a valuable tool for navigating the evolving AI regulatory landscape. Their success will ultimately depend on effective implementation, clear governance and meaningful cooperation between authorities.
The next step is therefore clear: turning the AI regulatory sandbox from a concept into a practical tool for responsible AI innovation.
Sources:
1 Regulation (EU) 2026/1744 of the European Parliament and of the Council of 8 July 2026 amending Regulations (EU) 2024/1689, (EU) 2018/1139 and (EU) 2023/1230 as regards the simplification of the implementation of harmonised rules on artificial intelligence
2 Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence
3 Article 57, paragraphe 3bis AI Act
4 Recital 24 AI Omnibus Regulation
5 Article 57, paragraphe 14 AI Omnibus Regulation
6 Recital 25 AI Omnibus Regulation; Article 58, paragraphe 1 (d) AI Omnibus Regulation
7 DOC 56 0856/043, 25 April 2025, pp. 21-22.; Belgian Chamber of Representatives, «Written question no. 44 by Mr Julien Matagne to the Minister of Self Employed Persons, Small and Medium-sized Enterprises and the Middle Classes: "Support for SMEs during the implementation of the AI Regulation" and answer of 5 September 2025», QRVA 56 023, September 2025.
