High-risk AI systems deployed in consequential domains must meet a growing set of legal obligations under the European AI Act, yet translating these obligations into concrete methods remains challenging. The European Commission has requested technical specifications in the form of harmonized standards, with terms such as accurate, robust, and trustworthy. Each of which, invoke processes and methods that vary greatly between disciplines and contexts.

In this session, we introduce and discuss the practical and technical challenges that arise when formalizing the AI Act’s legal obligations as concrete methods and measures. With a focus on discussing how these tools and methods are operationalized in practice, across industry, government and academia.

Tentative programme

Introduction and framing (10 min)

Invited talks (3 × 30 min)

Table discussions (2 × 30 min)

Plenary summary and closing (20 min)

Speakers’ list

Katerina Helena Vinther and Martin Skovbjerg Jensen
*Digitaliseringsstyrelsen*
Title: TBD.

Line K.H. Clemmensen and Ahcène Boubekki
Professor and Postdoc at UCPH
Title (attempt): Key insights from the YODA workshop

[TBC] Christopher Gyldenkærne
Postdoc at CAISA/UCPH
Title (attempt): Machine Learning, Healthcare, and Standards.

[TBC] Irina Alex Shklovski
Professor at UCPH
Title (attempt): Definitions, Shortcomings, and the Politics of AI Standards.

Level

Introductory: Suitable for beginners with little to no prior knowledge of the subject.

Organizers

• Anders Skaarup Johansen (Lead organizer), Postdoc (asjo@create.aau.dk), CAISA & AAU (TECH)
• Stella Graßhof, Associate Professor (stgr@itu.dk), IT University of Copenhagen (Data Science)
• Line H. Clemmensen, Professor (lkhc@math.ku.dk), KU (MATH)
• Ahcène Boubekki, Postdoc (ahcene.boubekki@math.ku.dk), KU (MATH).
• Senha Das, Associate Professor (send@dtu.dk), DTU