Algorithmic deterrence: A new strategic logic in the era of artificial intelligence

Abstract

The development of artificial intelligence represents a turning point in the modern understanding of deterrence strategy, shifting its focus from material power to information, perception, and the algorithmic management of actor behaviour. The classic concept of deterrence, as advocated by Thomas Schelling and Bernard Brodie, based on rational cost-benefit assessment, loses its predictive power in conditions of accelerated decision-making, autonomous systems, and algorithmic opacity. Instead of a stable "balance of fear", a state of dynamic instability arises, in which speed, information asymmetry and cognitive effects become key factors of strategic equilibrium. The paper's starting hypothesis is that artificial intelligence does not nullify deterrence but transforms it through a new cognitive and informational dimension: "algorithmic deterrence". Defence is no longer aimed exclusively at inflicting unacceptable damage, but at managing the adversary's perception, expectations, and behaviour in real time. Autonomous systems, machine learning, and predictive models enable the development of "intelligent deterrence" based on tailored strategic signals and algorithmic analysis of behaviour. Special attention is paid to the concepts of cognitive and information deterrence as modern forms of strategic communication. In this context, the paper argues that smaller states, such as Serbia, can develop asymmetric models of intelligent deterrence in the cyber and cognitive domains within the concept of total defence and digital sovereignty. In conclusion, the revitalisation of deterrence does not imply a return to classical models, but their transformation into a system of dynamic risk, information and perception management in the era of artificial intelligence.