On 14 October 2026 the European Commission published Commission Implementing Decision (EU) 2026/2274, which cites EN 18286:2026 as the first harmonised standard under the EU AI Act. The reference appears in the Official Journal and covers requirements for a quality management system that supports compliance with Article 17 and related provisions on technical documentation.
Scope and Content of EN 18286:2026
The standard was developed by CEN and CENELEC through Joint Technical Committee 21 following the standardisation request issued in June 2025. It was approved on 12 July 2026 and specifies a quality management system that addresses the full AI system lifecycle. Key elements include regulatory compliance strategy, design and development controls, testing and validation procedures, data management practices, post-market monitoring, serious incident reporting, record-keeping and accountability mechanisms.
EN 18286:2026 is intended to sit alongside existing frameworks such as ISO/IEC 42001 and ISO 9001. Informative annexes provide mapping tables that help organisations integrate the new requirements with their current management systems. The document is sector-neutral yet contains guidance on how it can be combined with industry-specific standards, for example ISO 13485 in medical device contexts.
Providers that implement the standard gain a presumption of conformity with the covered portions of the EU AI Act. Use of the standard remains voluntary; the presumption applies only to the requirements explicitly addressed in EN 18286:2026. The Commission has emphasised that further harmonised standards on risk management, data governance, accuracy, robustness and cybersecurity are still in development.
Timeline and Practical Steps for Enterprise Adoption
High-risk AI obligations under Chapter III of the EU AI Act apply from 2 December 2027 for standalone systems listed in Annex III and from 2 August 2028 for AI components embedded in products covered by Annex I. The early publication of the quality management standard gives organisations more than a year to prepare documented processes ahead of these dates.

Enterprises should begin by mapping existing quality and AI governance arrangements against the new standard. Gap analyses can identify where additional controls for post-market monitoring or incident reporting are required. Because the standard supports layered implementation, companies already certified to ISO 42001 or ISO 9001 can add the EU AI Act-specific elements without rebuilding their entire management system.
Legal and compliance teams will also need to review technical documentation practices, since the standard references the first sentence of Article 11(1). Documentation templates and record-keeping procedures should be updated to align with the requirements that trigger the presumption of conformity. Early adoption may reduce future audit and conformity-assessment costs once notified bodies begin evaluating high-risk systems.
Further guidance on transparency obligations for enterprise AI systems is available in the Commission’s earlier publication on the EU AI Act Guidelines Clarify Transparency Obligations for Enterprise AI Systems. Organisations tracking multiple regulatory streams can cross-reference the two documents when updating internal policies.
Next Steps and Remaining Standards
The Commission has described the publication of EN 18286:2026 as an important milestone. Additional harmonised standards requested under Article 40 are expected in the coming months. Providers that monitor the Official Journal and the work of JTC 21 will be able to incorporate new references as they become available.
Until the full set of standards is published, organisations should maintain flexibility in their compliance programmes. Documented risk assessments, data governance records and performance testing protocols developed now can be adjusted once the remaining standards are cited. This phased approach reduces the risk of rework while ensuring readiness for the 2027 and 2028 application dates.
Frequently asked questions
What does presumption of conformity mean under the new EU AI Act standard?
Organisations that implement EN 18286:2026 can rely on a legal presumption that they meet the covered parts of Article 17 and the first sentence of Article 11(1). The presumption is limited to those requirements and does not extend to other obligations in the regulation.
When must high-risk AI systems comply with the EU AI Act?
Standalone systems listed in Annex III must meet the obligations from 2 December 2027. AI components in products covered by Annex I face the requirements from 2 August 2028.
Can companies already certified to ISO 42001 use EN 18286:2026?
Yes. The standard is designed to be layered on top of ISO/IEC 42001 or ISO 9001. Mapping annexes help organisations integrate the additional EU AI Act controls without replacing existing certification.
Are other harmonised standards under the EU AI Act expected soon?
The Commission has requested further standards covering risk management, data governance, accuracy, robustness and cybersecurity. These have not yet been cited in the Official Journal.
By YasTech Editorial
Comments 1