OPINII Inteligenţa artificială împinsă cu forţa ı̂n vieţile noastre sau gardul proaspăt vopsit din faţa leopardului
OPINIONS Artificial intelligence forced into our lives or the freshly painted fence in front of the leopard

Summary/Abstract: Although Regulation (EU) 2024/1689 – commonly known as the AI Act – seeks to establish a unified legal framework for trustworthy, human‑centred artificial intelligence aligned with fundamental rights, a close examination of its internal structure reveals a profound structural tension between its normative discourse and contemporary technological reality. The provisions concerning prohibited practices and the strict obligations imposed on high‑risk systems are not merely preventive in nature; they are built upon an implicit acknowledgment of risks that are already visible. In other words, the AI Act does not prohibit hypothetical scenarios belonging to an uncertain future; it operates as a normative response to technological practices that already exist or are highly likely to emerge, for which technical evidence, commercial applications, or documented implementation plans are already available. This observation is central to understanding the ambivalent character of the regulation. On the one hand, it articulates a humanistic vision of technology, presenting itself as a guardian of human dignity, personal data protection, and social equity. On the other hand, by explicitly prohibiting subliminal manipulation, biometric classification based on sensitive traits, emotion recognition in contexts with power imbalances, social scoring systems, or the creation of facial‑recognition databases without consent, the AI Act indirectly confirms that such instruments are not mere theoretical possibilities, but technological realities already in circulation or in advanced stages of development. Thus, the prohibition is not an anticipatory act of prudence but a post‑factum reaction to a technological ecosystem that has evolved more rapidly than the regulatory framework intended to govern it. This article examines the dissonance between the regulation’s rhetoric of legal conformity and the concrete destabilizing potential of artificial intelligence. It argues that, rather than offering protection solely against speculative risks, the AI Act functions as a document that implicitly recognizes that AI systems are already capable of manipulating behavior, exploiting vulnerabilities, generating algorithmic discrimination, and operating beyond the understanding or control capacities of end‑users. In other words, what the regulation prohibits already exists as practice, technological trend, or economic intention among relevant actors. In this light, the AI Act may be interpreted as a normative „fence,” designed to inspire trust and convey an appearance of control, while concealing behind it an active „leopard” – a rapidly expanding technological ecosystem marked by opacity, power asymmetries, and substantial difficulties in ensuring effective oversight. This contrast between the legal form and the technological substance raises a fundamental issue of legitimacy and regulatory effectiveness: to what extent can a framework that acknowledges the existence of danger genuinely manage a phenomenon that has already surpassed the boundaries of traditional institutional control?

• Details
• Contents