Abstract

This article advances the Algorithmic Authoritarianism Hypothesis: that contemporary artificial intelligence does not generate authoritarianism ex nihilo, but functions as a powerful accelerator of its political, psychological, and institutional preconditions. Drawing primarily on Hannah Arendt’s concepts of natality, judgment, and the erosion of the common world, alongside Michel Foucault’s account of governmentality and Shoshana Zuboff’s analysis of instrumentarian power, the article argues that algorithmic governance reshapes the conditions under which democratic action remains possible. By rendering social behaviour predictable, optimisable, and anticipatable, AI systems displace political authorship from collective judgment toward technical modulation, producing a form of “soft authoritarianism” that preserves electoral forms while hollowing out substantive freedom. Using the United States as an illustrative case rather than an exceptional one, the article diagnoses a contemporary “digital Weimar” condition in which loneliness, epistemic fragmentation, and predictive control converge. Against this tendency, it reclaims political natality as the capacity to interrupt algorithmic recurrence through spontaneous action and judgment. The article concludes by outlining the contours of an “algorithmic humanism” oriented toward preserving unpredictability, plurality, and democratic agency under conditions of technological acceleration.