Delays in the Development of the EU AI Act Standards
The development of technical standards necessary for companies to demonstrate compliance with the EU’s AI Act is currently behind schedule, as reported by the main standardization bodies, CEN and CENELEC.
Background on the AI Act
In 2023, the European Commission requested that these organizations begin work on the technical standards that would allow manufacturers to prove that their products, services, or processes are compliant with the AI Act’s regulations. These standards are crucial for ensuring that AI applications are safe, trustworthy, and compliant with the established rules.
Current Status of Standards Development
Originally, the standards were expected to be completed by August 2025. However, CEN-CENELEC has indicated that, based on current project plans, this timeline will be extended into 2026. The organization emphasized the need to reflect the state of the art in AI technologies and to achieve consensus among European stakeholders during this process.
Process of Finalizing Standards
Once the initial drafts of the standards are ready later this year, they will undergo mandatory editing rounds, an assessment by the Commission, and will also require consultations and votes. This comprehensive procedure is projected to occupy much of 2025 and possibly extend into 2026 for some deliverables.
CEN-CENELEC has announced that they are implementing extraordinary measures to streamline the development timelines while maintaining close discussions with the AI Office.
Implications of the AI Act
The AI Act, which aims to regulate high-risk applications, became applicable in August 2024 and is gradually being implemented. The full enforcement of the act is slated for 2027.
By August 2025, member states are expected to establish national regulators responsible for overseeing compliance at the domestic level. These regulators will collaborate with the Commission’s AI Office, which operates within DG Connect.
Concerns from Regulatory Bodies
In a recent interview, a senior official from the Dutch data protection authority, which will oversee a portion of the AI Act in the Netherlands, expressed concerns about the ongoing process. The official highlighted that the standards are intended to provide certainty for companies and enable them to demonstrate compliance. However, they noted that there remains considerable work to be done before the standards are finalized, emphasizing that time is running out.
The development of these standards is critical for fostering trust and ensuring that AI technologies are implemented safely and responsibly within the European market.
