www.ciela.net
Author
Annotation
The article examines the administrative-law regime governing automated decision-making systems and artificial intelligence (AI) systems used in administrative proceedings and administrative justice. The author distinguishes rule-based machine systems from self-learning (machine-learning) systems. Their benefits and risks are illustrated through foreign and Bulgarian practice, including China's internet courts, SyRI, iBorderCtrl, the Sofia Municipality kindergarten admission system and electronic traffic fines. The article analyses Regulation (EU) 2024/1689 (the AI Act): its principles, its risk-based approach (prohibited practices, high-risk systems and systems subject to transparency obligations), ex-ante and ex-post control, and the protection of affected persons. It argues that acts of notifying authorities and decisions based on high-risk AI systems are individual administrative acts that can be appealed under the Administrative Procedure Code. The author criticises the AI Act as overly detailed and restrictive, saying it leads to over-regulation and undermines the EU's competitiveness. He therefore recommends that Bulgaria should not adopt a separate AI law and should apply the Administrative Procedure Code instead.
Keywords
artificial intelligence; automated decision-making systems; AI Act; high-risk AI systems; administrative procedure
