The European Parliament has published the draft report as well as released three batches of amendements from the LIBE and IMCO committees (co-leads) on the Digital Omnibus on AI proposal.

‍

SUMMARY:

• The European Parliament has published the draft report and first batch of amendements from the LIBE and IMCO committees (co-leads) on the Digital Omnibus on AI proposal.
• A key positive amendment, in line with FESI’s position shared with the rapporteurs, is the replacement of the Commission’s proposal to link the application date to the adoption of supporting tools by the Commission with fixed application deadlines:   2 December 2027 for Annex III systems and 2 August 2028 for Annex I systems. Althoug we asked for a single deadline for both annexes, it still reflects FESI’s call for legal certainty and the removal of conditionality.
• The committees’ vote should take place on 18 March, followed by plenary on 25 March (tbc). Trilogue negotiations are expected in March–April, with final plenary adoption planned for June. The objective remains entry into force of the amended parts of the AI Act on 1 August, one day before the initial application of high-risk AI obligations under Annex III.

MORE INFORMATION:

• ‍Draft report: The draft report has been drafted by MEP Arba Kokalari (IMCO rapporteur) and Michael McNamara (LIBE), which whom FESI has already shared its position paper and amendment proposal.
• The Co-Rapporteurs consider that a postponement of the application date of the AI Act is necessary, considering the delayed preparation of standards, which are necessary to support compliance, as well as the delayed guidelines, governance and conformity assessment frameworks.
• In order to ensure legal certainty and predictability, the report suggests to replace the Commission’s proposal of linking the date of application to a decision by the Commission with a set timeline of 2 December 2027 for Annex III systems and 2 August 2028 for Annex I systems (see amendment 6).
• This is fully aligned with FESI’s position and amendment proposal, which explicitly called for a fixed deadline instead of Commission conditionality.

In addition, the report states that the Commission should ensure that measures in support of compliance are in place in due time to avoid further application delays (see amendment 7).

This also mirrors FESI’s request that all remaining secondary legislation and guidance be adopted without undue delay to guarantee a workable implementation framework.

Besides the postponement, the Co-Rapporteurs have introduced some targeted adjustments to the Commission’s proposal related to AI literacy (see notably amendments 8 & 9), processing of special categories of personal data for bias detection and mitigation, notified bodies, cybersecurity and sandboxes.

Amendements

A first package of amendements (171 - 450) have also been published. Key points for FESI include:

• Postponement of the application timeline: Axel Voss (EPP, Germany), with whom FESI participated in a stakeholders roundtable in early February, tabled amendements (23 - 26) removing the conditionality and aligning with the Council wording of 23 January to introduce fixed deadlines: 2 December 2027 for Annex III systems and 2 August 2028 for Annex I systems
• Registration of non-high-risk AI systems: Several amendments delete the obligation for providers who considers that an AI system referred to in Annex III is not high-risk shall to document their assessment. They also delete the obligation to provide documentation automatically upon request to national  authorities (see AMs 198 to 202). Other amendments ensure authorities only request what is necessary to verify the Article 6(3) self-assessment (AMs 203 and 204). Another one introduce the requirement that requests be “reasonable” (AM 207). These amendments are in line with FESI’s call for proportionality.
• Implementing Acts vs Guidance (Articles 50, 56, 72): several amendment delete the Commission’s proposal to replace implementing acts by guidance such as codes of conduct, stating that implementing acts are vital to prevent divergent national approaches and that removing them shifts the burden to ad-hoc interpretations, increasing uncertainty and costs for AI operators (AM 251 and 252).     

‍

In the second package of amendments (318 - 450) , key points for FESI include the application timelines for high-risk AI obligations.

The majority of these amendments aim to replace the conditional mechanism initially proposed by the Commission with fixed timelines (see from AM 396). The shortest timeline proposed for Annex III is 1st December 2026 (AM 402).

Importantly, two amendments propose a harmonized date of 2 August 2028 for both Annexes, in line with FESI’s recommendations.

Among these, Amendment 399 exactly mirrors the text FESI drafted and circulated to MEPs, requesting the 2 August 2028 deadline and that the Commission shall adopt all remaining pieces of secondary legislation and guidelines without undue delay.

‍

In the third package of amendments (25 - 170), the most relevant focus on AI literacy, the registration of non-high-risk AI systems and timelines.

AI Literacy:

1. Amendment 42: Replaces the obligation for the Commission and Member States to encourage providers and deployers to ensure AI literacy with a requirement for the Commission and the European Data Protection Board to issue guidance on practical implementation.
2. Amendments 43, 136, 137: Maintain or impose a general obligation on providers and deployers to ensure AI literacy of their staff.
3. Amendment 44: Replaces “encourage” with “provide support, if needed, to providers and deployers” for AI literacy. It clarifies role of Commission/Member States without creating rigid obligations.
4. Amendment 45 & 139: Emphasize the need for Member States and the European Commission to provide guidance and support to providers and deployers for AI literacy, especially for SMEs, instead of a general obligation on providers and deployers.
5. Amendments 126-128, 130: Delete the EC’s proposal on AI literacy (Article 4).
6. Amendment 129: Requires guidance from the Commission and the EDPB on practical implementation of AI literacy obligations.

Registration of Non-High-Risk AI Systems

1. Amendments 60–65: Delete Recital 9, which exempted certain non-high-risk AI systems from registration if no significant harm occurs.
2. Amendment 66: Maintains mandatory registration for non-high-risk systems but allows simplified procedures.

Timeline and Transitional Provisions

1. Amendments 84–87: Delete the 6-month transition period for systems already on the market before 2 August 2026.
2. Amendment 95: Timeline should only be extended for obligations dependent on technical standards; other provisions remain as originally scheduled (2 August 2026).
3. Amendment 96: Aligns implementation of Sections 1–3 of Chapter III for high-risk AI systems to 2 August 2028, ensuring harmonized standards are available in advance.
4. Amendment 97: Defers obligations when necessary guidelines/standards are missing.

The final trilogue agreement is expected to be voted in plenary in June, as the Parliament aims for publication in the Official EU journal in July and entry into force on 1^st August.