The development and uptake of Artificial intelligence (AI) technologies across various applications and policy domains is increasingly taking up the scale. In response to this development, in 2020 CEN-CENELEC established a Focus Group on AI and published a roadmap for AI standardisation and in 2021 created a Joint Technical Committee on Artificial Intelligence (CEN-CENELEC JTC21). In April 2021, the European Commission published a legislative proposal on AI, which introduces harmonised horizontal requirements for a limited number of clearly defined AI systems that pose high risks to safety and fundamental rights with the key objective of ensuring that AI systems used or put into services on EU internal market are safe and trustworthy.

The AI Act proposal is a horizontal legislation that follows a New-Legislative Framework (NLF) approach. In other words, high-level essential requirements are provided in the main legal act, while technical operationalisation is ensured using harmonised standards. European harmonised standards developed by Euopean Standardisation Organisations (ESOs), and approved by the Commission, are therefore essential tools for the implementation of the AI Act and for the presumption of conformity that should facilitate the development of trustworthy AI technologies and trust in those technologies by citizens.

In order to facilitate the development of AI standards and availability of those standards already before the date of application of the AI Act, the Commission has recently issued a Standardisation Request (SR) which requests ESOs to work on 10 areas of AI standardisation deliverables, including: data governance and data quality, record keeping, provision of information and transparency, robustness, accuracy, cybersecurity, human oversight, risk management and testing, conformity assessment and quality management.

Notwithstanding the general horizontal nature of the AI Act and the Standardisation Request, European standardisation deliverables may also provide, if necessary, specifications that concern certain specific AI systems (use cases) or sectors (vertical specifications), in consideration of the intended purpose and/or context of use of those systems.

In this context, to support the current AI standardisation activities at European level, the European Commission and CEN-CENELEC JTC21 have decided to jointly organise a workshop on the interplay between horizontal and vertical standardisation deliverables. This workshop intends to:

1. present the approach to AI standardisation proposed by the European Commission in the AI Act and the AI Standardisation Request,
2. discuss options for the involvement of sectoral technical committees in the development of horizontal AI standardisation deliverables, and
3. present case studies of early AI standardisation activities carried out in specific sectors.