The AI Act, Artificial Intelligence, and Public Administration: Prospects for Use and Misuse in Social Services and Administrative Activities

This study examines the regulatory, organizational, and social implications of artificial intelligence in public administration. It distinguishes traditional algorithmic systems from generative AI and considers their growing role in data management and administrative decision-making. The analysis focuses on two cases. In France, an AI system used to identify undeclared private swimming pools generated errors and risks of unequal treatment. In the Netherlands, automated systems for allocating child benefits caused serious economic and social harm to vulnerable families, raising concerns about discrimination and violations of fundamental rights. The paper then assesses the GDPR and the AI Act. Although these frameworks establish important principles of privacy, transparency, and accountability, they leave gaps concerning AI use in administrative procedures, especially regarding private-sector involvement and citizen participation. The study concludes that regulation alone cannot guarantee the responsible use of AI in public administration. Binding rules must be combined with civil servant training, citizens’ digital literacy, and applied research to create an administrative ecosystem capable of using AI safely, fairly, and in full respect of fundamental rights.