Nicolae Iorga street, 2/1,
Tîrgu-Mureș
postal code: 540088
telephone:0265-250102;
0265-255554
e-mail: redactie@revcurentjur.ro
Year: 2026
Issue: 2
Volume: 105
Author Names: Dragoș Alexandru MAIOR
Abstract: Inevitably, the expansion of artificial intelligence (AI) technologies was bound to intersect with the realm of national security. Driven by this objective, the prospect of unprecedented predictive surveillance and analytical processing capabilities is undeniably compelling. Yet, while the national security exemption is enshrined at the European Union level, finding its explicit regulation within the AI Act itself, its very nature demands a strict interpretation. Addressing these emerging challenges, although government agencies might deploy AI-based technical solutions that qualify as high-risk systems, the AI Act mandates an ex-ante oversight mechanism. This paper traces and dissects this legal requirement through the lens of Legislative Bill 565/2025 amending Law No. 51/1991 regarding Romania's national security, a legal instrument empowering intelligence agencies to conduct mass processing of personal data via automated means. Anchored in this legislative draft, which has tacitly passed the first chamber of the Parliament, the present study argues that when dealing with this specific class of technology, traditional parliamentary oversight proves inadequate. Due to its fundamentally reactive and ex-post nature, such oversight becomes ineffective and merely illusory when confronted with the operational opacity inherent to these systems.
DOI: https://doi.org/10.62838/cjjc-2024-0081
Pages: 59-70