Simplification of the implementation of harmonised rules on artificial intelligence (Digital Omnibus on AI)
TA-10-2026-0098 · P10_TA(2026)0098 · PE782.530
- Adopted in plenary on · Adopted 569–45–23 · See the decision →
- Full procedure 2025/0359
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.
Parliament's amended version of the proposal to simplify the AI Act (Regulation (EU) 2024/1689) and related rules. It adjusts definitions, obligations, and procedures to reduce burdens, especially for SMEs and SMCs, while maintaining protections.¶¶¶ It introduces a ban on AI that generates non-consensual sexual deepfakes, clarifies safety functions, simplifies registration and conformity assessment, and extends AI Office supervision to all systems based on general-purpose AI models by the same provider.¶¶ It also adjusts application dates, removes certain Commission powers, and moves sectoral legislation from Annex I Section A to Section B, affecting conformity assessment.¶¶¶
Key points
- Defines SMEs and SMCs, extending certain measures for small-scale providers to SMCs where appropriate, while safeguarding the AI Act's objectives.¶
- Changes AI literacy obligation: providers and deployers must support AI literacy, but no guarantee of a specific level; Commission to issue guidance and support, especially for SMEs.¶¶¶
- Allows exceptional processing of special categories of personal data for bias detection and correction in high-risk AI, under strict safeguards.¶¶¶
- Bans AI systems that generate non-consensual sexual deepfakes, with exceptions for providers with effective safety measures and for developing capabilities.¶¶
- Clarifies that an AI system is high-risk only if it is a safety component necessary for compliance with Union safety requirements; excludes non-safety functionalities.¶¶¶
- Simplifies registration of AI systems under Article 6(3) in the EU database, but providers must still document assessments.¶
- Extends real-world testing to high-risk AI systems covered by Annex I, with voluntary agreements for Section B, and allows AI Office to establish an EU-level sandbox.¶¶¶
- Clarifies AI Office supervision: it covers all AI systems based on general-purpose AI models by the same provider or group, excluding certain products and Annex III point 2 systems.¶¶
- Requires the Commission to ensure pre-market conformity assessments for AI systems under AI Office supervision, and to maintain records of communications.¶¶
- Strengthens cooperation between market surveillance and fundamental rights authorities, with mutual assistance and access to information without undue delay.¶¶¶
- Sets a 3-month transitional period for providers of generative AI systems already on the market to comply with marking obligations by 2 November 2026.¶¶
- Removes certain Commission implementing powers, moves sectoral legislation from Annex I Section A to Section B, and adjusts application dates for high-risk obligations.¶¶¶
Who is affected
- Providers and deployers of AI systems, especially SMEs and SMCs, benefit from simplified obligations and reduced administrative burden.¶¶¶
- Providers of AI systems that generate sexual deepfakes face a ban, unless they have effective safety measures.¶¶
- The AI Office gains new supervisory powers and must be allocated adequate resources.¶¶¶
- National authorities and bodies protecting fundamental rights get enhanced cooperation and access to information.¶¶¶
- Manufacturers of products covered by sectoral legislation moved to Annex I Section B see changes in conformity assessment.¶¶
Figures and deadlines
- 99.8% of all Union companies are small and medium-sized enterprises.¶
- Transitional period of 3 months for providers of generative AI systems already on the market before 2 August 2026.¶
- Providers must comply with Article 50(2) by 2 November 2026.¶
- Commission to adopt guidance on post-market monitoring plan by 2 February 2027.¶