AI Act update: postponement of high-risk obligations and clarification of transparency requirements
On 7 May 2026, the Council of the European Union and the European Parliament reached a provisional agreement on a series of amendments to the AI Act as part of the so-called Omnibus VII legislative package. These amendments form part of the broader European simplification agenda, through which the European Commission aims to reduce administrative burdens for businesses and strengthen the competitiveness of the European economy.
Postponement for high-hisk AI systems
One of the key amendments concerns the postponement of several obligations applicable to high-risk AI systems. These extensions reflect the challenges associated with operationalising various provisions of the AI Act, particularly for high-risk AI systems that require conformity assessments, documentation, and external evaluation.
The Commission has proposed postponing the application of the rules for high-risk AI systems by up to 16 months. Specifically, the deadline for standalone high-risk AI systems is postponed from 2 August 2026 to 2 December 2027. For high-risk AI systems integrated into regulated products, the deadline is extended from 2 August 2027 to 2 August 2028.
In addition, the deadline for Member States to establish AI regulatory sandboxes is postponed until 2 August 2027. Certain exemptions for SMEs are also extended to small mid-cap companies.
Expansion of prohibited AI practices
In addition to the amendments outlined above, the agreement also includes a substantive expansion of the prohibited AI practices set out in Article 5 of the AI Act. A new explicit prohibition is introduced on the use of AI systems to generate or manipulate non-consensual intimate material, as well as child sexual abuse material ("AI nudifier" systems). This provision specifically targets applications that use deepfake technology to create sexual or explicit images of individuals without their consent. The prohibition will enter into force on 2 December 2026 and highlights the growing focus of the European legislator on protecting fundamental rights and human dignity in the context of generative AI.
The political agreement must still be translated into formal amendments to the AI Act. However, the European institutions have indicated their intention to finalise this process in the short term and, in any event, before 2 August 2026.
New guidance on transparency
On 8 May 2026, the European Commission also published draft guidance on the implementation of the transparency obligations for certain AI systems under Article 50 of the AI Act. These guidelines are intended to support providers, deployers, and supervisory authorities in ensuring a consistent and uniform application of the transparency rules for certain AI systems, including systems that generate AI-created content or interact with natural persons. Stakeholders were invited to provide feedback until 3 June 2026. In parallel, a code of practice for the marking and labelling of AI-generated content was developed.
Preparation remains essential
Although several obligations have been postponed, it remains clear that businesses will need to prepare in a timely manner for the implementation of the AI Act. Organisations that develop, integrate, or use AI systems would be well advised to map their governance, compliance, and risk management processes today. The recent amendments demonstrate that European AI regulation continues to evolve and become more refined, while at the same time taking increasingly concrete shape.
Do you have questions about the impact of the AI Act on your organisation, or would you like support in preparing for the new obligations? Moore Law closely monitors developments and is happy to assist businesses with the legal and practical implementation of AI regulation.
Key milestones of the AI Act
The timeline below provides an overview of the main milestones and application dates of the AI Act, taking into account the recently proposed amendments included in the Omnibus VII package.
| Date | Milestone |
| 1 August 2024 | Entry into force of the AI Regulation |
| 2 February 2025 | Application of general provisions and the prohibition of certain AI systems, as well as AI literacy obligations (Chapters I and II) |
| 2 August 2025 | Application of rules for General-Purpose AI (GPAI) models and designation of competent supervisory authorities |
| 2 August 2026 | Article 50 (transparency) enters into force as planned, with a proposed four-month transition period for existing systems under Article 50(2) |
| 2 December 2026 | AI systems that generate or manipulate content and that were placed on the market before 2 August 2026 must be marked in a machine-readable format and be recognisable as artificially generated or manipulated. Entry into force of the newly proposed prohibition on AI nudifier systems (amendment to Article 5) |
| 2 December 2027 | Application of rules for standalone high-risk AI systems (Annex III) |
| 2 August 2027 | Obligation for Member States to establish at least one AI regulatory sandbox |
| 2 August 2028 | Application of rules for high-risk AI systems integrated into regulated products (Annex I) |
