The GDPR’s data protection impact assessment framework applied to deep-learning AI systems: a noteworthy asset for the conduction of algorithmic impact assessments

Strowel, AlainCeysens, NicolasNicolasCeysens2025-02-042025-02-042023https://dial-mem.test.bib.ucl.ac.be/handle/123456789/35627The subject of artificial intelligence has known a growing interest this last decade, with impressive developments only recently. This is not without raising concerns related to privacy and accountability. This thesis explores how this reality challenges the existing data protection framework provided by the GDPR, specifically considering the characteristics of of deep-learning AI systems. It focuses on the way GDPR envisages this reality through the existing data protection impact assessment, required in certain cases by Article 35 of the Regulation. The thesis also addresses the interest a data protection impact assessment can have in the conduction of an algorithmic impact assessment, in light of the upcoming European framework for AI.RGPDDPIAAlgorithmic impact assessmentsArtificial intelligenceDeep-LearningThe GDPR’s data protection impact assessment framework applied to deep-learning AI systems: a noteworthy asset for the conduction of algorithmic impact assessmentstext::thesis::master thesisthesis:42864