New DIGITALEUROPE paper responds to criticism of AI Act Annex I changes
A new document from DIGITALEUROPE seeks to counter criticism of proposed changes to the EU AI Act. The paper argues that merging parts of Annex I would simplify compliance without weakening safeguards, as debate continues ahead of trilogue negotiations.
DIGITALEUROPE has published a detailed Q&A responding to criticism of the proposed merger of Sections A and B of Annex I of the EU AI Act, positioning the reform as a procedural adjustment rather than a reduction in regulatory safeguards.
The document, released on 9 April 2026, addresses concerns raised by certification bodies, civil society organisations, trade unions, and parts of the European Commission. According to the paper, several criticisms are based on “factual inaccuracies” that risk influencing ongoing negotiations on the AI Act.
At the centre of the debate is how high-risk AI systems embedded in regulated products, such as medical devices and machinery, should be assessed. Under the current framework, these products must comply with both sector-specific legislation and the AI Act in parallel. DIGITALEUROPE argues that this creates duplication and uncertainty, requiring manufacturers to undergo two separate conformity assessments for the same product.
The proposed merger would shift all such products to a model where AI requirements are integrated into sectoral legislation. The organisation states that this would create a single compliance pathway, while maintaining all existing high-risk obligations under the AI Act.
The Q&A directly challenges claims that the merger would weaken protections or exempt certain sectors. It states that no products would be removed from the scope of the AI Act and that requirements related to risk management, transparency, and human oversight would remain fully applicable.
DIGITALEUROPE also disputes arguments that the reform would delay enforcement. The document argues that implementation challenges already exist under the current system, including gaps in harmonised standards and limited capacity among notified bodies. It presents the merger as a way to address these structural issues rather than introduce new delays.
On governance, the organisation rejects concerns that embedding AI rules into sectoral frameworks would fragment the AI Act’s horizontal approach. It argues that a similar model already exists within the legislation and that extending it would not alter the overall structure.
The paper also responds to accusations that the proposal reflects industry lobbying aimed at deregulation. DIGITALEUROPE states that the changes are procedural and intended to reduce administrative burden, not to remove safety requirements. It cites compliance costs and operational challenges faced by manufacturers under the current dual system.
The publication comes shortly after the European Parliament adopted its position on the Annex I merger in March 2026, giving political momentum to the proposal ahead of negotiations with the Council and the Commission.
The Q&A reflects the broader tension in the AI Act debate between maintaining a uniform regulatory framework and adapting oversight to sector-specific realities.
