Opportunity to Improve the AI Act – Will the EU Seize the Moment?
The European Union seeks to simplify its extensive digital regulatory framework and to ensure that innovation is not impeded by unnecessarily complex or burdensome rules. As part of this effort, the European Commission has presented a Digital Omnibus on AI, which amends and streamlines certain elements of the EU Artificial Intelligence Act (AI Act). The objective is to ensure that the AI Act is practically applicable, legally clear, and proportionate, while maintaining a high level of protection for fundamental rights and consumer safety.
Within the EU, there is an increasing awareness that excessive regulatory burden undermines competitiveness. Against this background, the Commission proposes, inter alia, extended timelines for the application of provisions relating to high‑risk AI systems, more flexible rules governing data processing in the development of AI systems, and simplified compliance requirements for smaller businesses. The overall aim is to reduce administrative costs and to create more favourable conditions for innovation.
The Confederation of Swedish Enterprise main positions are
The date of application of the rules governing high‑risk AI systems is highly time‑critical and should therefore be addressed separately.
An insufficiently clear delineation of what constitutes high‑risk AI creates significant legal uncertainty. Regulation should be based on actual risks, rather than on specific technologies or entire sectors.
The requirement relating to AI literacy should be removed, as legislative intervention is not necessary for companies to ensure appropriate handling and use of AI systems within their organisations.
The possibility to process special categories of personal data is necessary and practically justified, for example for the purpose of bias mitigation; however, the current proposal is not sufficient to achieve a substantive effect in practice.
The ability to test AI systems in real‑world conditions is essential for innovation, effective risk management, and system adaptation, and should therefore apply to all AI systems, not only those tested within regulatory sandboxes.
Administrative simplifications are welcome, but regulatory requirements must be designed in such a way that all undertakings, regardless of size, are able to comply. This is essential to ensure a level playing field and to avoid growth‑inhibiting barriers. Exemptions may also create challenges within supply chains, particularly where subcontractors are subject to simplified documentation requirements.
Overlapping regulatory requirements—such as those relating to notified bodies and documentation obligations across different legislative instruments—need to be addressed and streamlined.
The Confederation of Swedish Enterprise considers the EU’s Digital Omnibus on AI to be a first step towards a more predictable and less bureaucratic regulatory framework capable of strengthening competitiveness. For Swedish companies, it is crucial that the EU now takes this opportunity to address key shortcomings in the AI Act and to ensure that it is proportionate, technology‑neutral, and based on clear, risk‑based requirements and timelines that make compliance feasible in practice.
