Legal Protection against Defective Algorithmic Administrative Decisions

Abstract

This study addresses the topic of "Legal Protection against Defective Algorithmic Administrative Decisions," aiming to evaluate the adequacy and guarantees of current legal, regulatory, and judicial frameworks in providing effective protection for individuals against the risks and flaws of fully automated decisions. Utilizing a comparative analytical approach (drawing insights from French, Italian, and European Union laws), the research applies these concepts pragmatically to the Libyan administrative and financial context by examining active systems (the financial compliance system "Ratbak Lahzi" at the Ministry of Finance, the foreign currency reservation platform at the Central Bank of Libya, and the wife and children grant system at the Ministry of Social Affairs). The study reveals a profound regulatory gap and legislative vacuum in Libya, which breeds risks of "executive improvisation" and the tyranny of the "Black Box" that conceals decision logic from both litigants and judges, leading to a crisis in the burden of proof. The study recommends the urgent need to codify AI within the administrative domain, mandate human-in-the-loop intervention, enforce qualified transparency and audit logs, activate the roles of sovereign oversight bodies (the Administrative Control Authority and the Audit Bureau), and adopt strict liability rules based on the risk theory and the combination of liabilities to compensate affected individuals.