The model component: accurate and secure inference

Bjørn Aslak Juliussen

Chapter 4 in Compliance by Design in AI Systems, 2026, pp 131-213 from Edward Elgar Publishing

Abstract: This chapter analyses the legal requirements applicable to deployed AI systems under the GDPR, the AI Act, and the Cyber Resilience Act (CRA). It highlights how fundamental principles for processing personal data in the GDPR, particularly accuracy, purpose limitation, and data minimisation, apply in trained and deployed AI systems. The chapter outlines the requirements for high-risk and general-purpose AI systems. CRA requirements are discussed in relation to AI-specific vulnerabilities such as model inversion and inference attacks, emphasising the need for targeted mitigation strategies. The chapter also explores the evolving interpretation of Article 22 GDPR following the Schufa judgement, the right to explanation under both the GDPR and the AI Act, and how explainable AI (XAI) methods can support compliance. Finally, it considers how MLOps practices can operationalise legal requirements by embedding accuracy, security, and purpose limitation in deployed AI models.

Keywords: Personal Data Processing; Deployed Models; Accuracy; Purpose Limitation; Explainability; XAI (search for similar items in EconPapers)
Date: 2026
ISBN: 9781035390724
References: Add references at CitEc
Citations:

Downloads: (external link)
https://www.elgaronline.com/doi/10.4337/9781035390731.00012 (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:elg:eechap:25422_4

Ordering information: This item can be ordered from
http://www.e-elgar.com

Access Statistics for this chapter

More chapters in Chapters from Edward Elgar Publishing
Bibliographic data for series maintained by Jack Sweeney ().