On the proposal for a regulation of the European Parliament and of the Council amending Regulations (EU) 2024/1689 and (EU) 2018/1139 as regards the simplification of the implementation of harmonised rules on artificial intelligence (Digital Omnibus on AI)
JURI-AD-784179
Committee on Legal Affairs · Rapporteur: Sergey Lagodinsky
Written by a language model from the full text only; every figure comes from the text. The paragraph links open the paragraph a statement rests on — check the text itself before relying on it.
This committee opinion proposes amendments to the Commission's proposal to simplify the AI Act, focusing on proportionality, clarifying definitions, and reducing burdens for SMEs and SMCs.¶¶¶ It adds prohibitions on AI generating non-consensual sexual content and child sexual abuse material, and clarifies that AI agents fall under the AI system definition.¶¶¶ It adjusts high-risk classification, extends transition periods for micro-enterprises, and requires the Commission to adopt common specifications by 2 December 2027 if standards are delayed.¶¶¶
Committee position. The Committee on Legal Affairs proposes amendments to the Commission's proposal, focusing on proportionality, clarifying definitions, and reducing burdens for SMEs and SMCs, while adding prohibitions on certain AI practices.¶¶¶
Key points
- Obligations under the AI Act should be implemented proportionately, considering the nature, scale and complexity of activities.¶
- The simplification should not alter the AI Act's scope or introduce new legal categories, and should not prejudge future policy on autonomous systems.¶
- Extending SME-specific rules to SMCs should be exceptional and only for specific provisions, to avoid harming micro-enterprises.¶¶
- AI agents are clarified as falling within the AI system definition, and the Commission and Member States should provide information assistance to SMEs and startups on risk classification.¶¶
- The AI literacy obligation is changed from mandatory to encouraging, with Member States and the Commission supporting providers and deployers.¶¶
- AI systems that generate non-consensual sexual or nude content, or child sexual abuse material, are added to prohibited practices.¶¶
- Bias detection and correction using special categories of personal data is allowed for high-risk and other systems under strict conditions, without creating an obligation.¶¶¶
- High-risk classification is clarified: an AI system is high-risk only if it is a safety component and its functionality impacts safety; lack of harmonised standards alone does not make it high-risk.¶¶
- Micro-enterprises using high-risk AI internally for process optimisation are exempt from certain obligations, with simplified documentation, under conditions.¶
- Providers that are SMEs or SMCs receive tailored guidance from national authorities for technical documentation and quality management.¶
- The Commission must issue standardisation requests by 2 December 2027 and adopt common specifications by that date if standards are lacking, with Council authorisation for urgent cases.¶¶¶
- Micro-enterprises get an additional 24-month transition period for high-risk AI obligations, with a non-sanctioning monitoring regime and a Commission report due within 36 months.¶
Who is affected
- Providers and deployers of AI systems, especially SMEs and SMCs, face adjusted obligations and support measures.¶¶¶
- Micro-enterprises using AI internally for process optimisation benefit from exemptions and extended transition periods.¶¶
- National authorities gain direct access to technical documentation for fundamental rights enforcement.¶
- The AI Office and national authorities cooperate on sandboxes and enforcement, with SMEs and public administrations supported.¶¶¶
Figures and deadlines
- 2 December 2027: deadline for the Commission to adopt common specifications and issue standardisation requests.¶¶¶
- 24 months: additional transition period for micro-enterprises for high-risk AI obligations.¶
- 36 months: deadline for the Commission to report on micro-enterprises' compliance capacity.
- 31 August 2029: date by which providers of high-risk AI systems intended for public authorities must comply.¶