When Algorithms Replace Parties: AI Surveillance and Elite Power-Sharing in Autocracies

Authoritarian rulers in the Global South and East govern under acute uncertainty. Unlike their counterparts in consolidated autocracies of the twentieth century, contemporary regimes in Africa, the Middle East, post-Soviet Eurasia, and Southeast Asia frequently operate in contexts of weak institutions, fragmented elites, and volatile state capacity. Classic political economy theories of authoritarianism argue that dictators mitigate these challenges by relying on political parties, elections, and bureaucratic organizations to reduce information asymmetries, monitor elite behavior, and sustain ruling coalitions through institutionalized power-sharing. These institutions historically served as mechanisms to detect disloyalty, coordinate elite behavior, and gather information about societal preferences. This paper argues that the diffusion of AI-enabled surveillance technologies (such as facial recognition and predictive policing) fundamentally transforms this equilibrium in contemporary autocracies. In contrast to earlier forms of coercive monitoring, AI-based systems provide rulers with direct, fine-grained, and continuously updated information about both citizens and elites at unprecedented scale and declining marginal cost. This shift is particularly consequential in the Global South and East, where regimes often face acute informational deficits and where digital infrastructure has expanded rapidly without commensurate institutional checks, frequently facilitated by transnational technology transfer and security cooperation. The argument proceeds in three steps. First, by dramatically improving rulers’ ability to monitor both mass and elite behavior, AI surveillance increases informational asymmetries between the dictator and members of the ruling coalition. Elites, aware of their heightened visibility and the risks of being labeled as potential plotters, engage in preference falsification and reduce horizontal communication, weakening their capacity to coordinate and bargain collectively. This dynamic reduces elite divisions and increases intra-elite cohesion around the ruler (H1). Second, AI-based surveillance substitutes for traditional information-gathering institutions. As rulers gain direct access to local-level data on social behavior and political attitudes, the functional value of strong party organizations and their embeddedness in social organizations declines, leading to the erosion of local party branches and weakening linkages with civil society networks (H2a–H2b). Third, the centralization of informational control at the apex of the regime fosters a shift toward personalist rule. Surveillance capacities become politically instrumentalized through techno-populist narratives that portray the leader as a technologically competent guarantor of order, thus reorienting legitimation away from institutions and toward the ruler’s persona (H3). Empirically, the paper draws on the AI & Big Data Global Surveillance Index to track the diffusion of AI-enabled surveillance technologies across 65 authoritarian regimes in the Global South and East between 2000 and 2023. These data are combined with party organization, elite cohesion, and personalism indicators from V-Party. Using panel regressions with regime and year fixed effects, the results show that the adoption of AI surveillance is associated with increased elite cohesion and personalization of power, alongside systematic erosion of local party structures and societal linkages. These findings shed light on how digital technologies are reshaping the political economy of authoritarianism beyond the Global North.