Invisible threats: Looking back to move forward with AI - The multidimensional impact of AI on organizational security and human agency

As artificial intelligence systems increasingly permeate all domains of human activity - from healthcare and education to employment and governance - new forms of security risks emerge, many of which remain insufficiently recognized. This paper explores the multidimensional infiltration of AI into socio-organizational systems, analyzing not only the technological risks, but also the psychological, organizational, and epistemological threats posed by unchecked automation and opaque algorithmic decision-making. Drawing from human resource management (HRM) practices in digitally transformed organizations, the paper identifies early warning signals of AI misuse: diminished human agency, erosion of professional autonomy, over-reliance on predictive analytics, and loss of trust among stakeholders. By extending these findings, the study outlines future development scenarios, ranging from responsible integration of AI grounded in stakeholder dialogue to dystopian trajectories marked by dehumanization and social fragmentation. The final section presents strategic recommendations for promoting responsible AI use, emphasizing the need for transdisciplinary education, critical digital literacy, and anticipatory governance. In shaping the next generation of experts, the role of academia and research institutions becomes vital in embedding ethical reflexivity, systemic thinking, and human-centric values into the very design of AI-driven futures.