Waltzing into uncertainty: AI in nuclear decision making and the challenge of divergent deterrence logics

Zatsepina, L ORCID: 0000-0001-7321-3232 (2026) Waltzing into uncertainty: AI in nuclear decision making and the challenge of divergent deterrence logics. Cambridge Forum on AI: Law and Governance, 1. pp. 1-15. ISSN 3033-3733

Preview

Text waltzing-into-uncertainty-ai-in-nuclear-decision-making-and-the-challenge-of-divergent-deterrence-logics.pdf - Published Version Available under License Creative Commons Attribution. Download (282kB) | Preview

Abstract

This article critically examines the integration of artificial intelligence (AI) into nuclear decision-making processes and its implications for deterrence strategies in the Third Nuclear Age. While realist deterrence logic assumes that the threat of mutual destruction compels rational actors to act cautiously, AI disrupts this by adding speed, opacity and algorithmic biases to decision-making processes. The article focuses on the case of Russia to explore how different understandings of deterrence among nuclear powers could increase the risk of misperceptions and inadvertent escalation in an AI-influenced strategic environment. I argue that AI does not operate in a conceptual vacuum: the effects of its integration depend on the strategic assumptions guiding its use. As such, divergent interpretations of deterrence may render AI-supported decision making more unpredictable, particularly in high-stakes nuclear contexts. I also consider how these risks intersect with broader arms race dynamics. Specifically, the pursuit of AI-enabled capabilities by global powers is not only accelerating military modernisation but also intensifying the security dilemma, as each side fears falling behind. In light of these challenges, this article calls for greater attention to conceptual divergence in deterrence thinking, alongside transparency protocols and confidence-building measures aimed at mitigating misunderstandings and promoting stability in an increasingly automated military landscape.

Item Type:	Article
Uncontrolled Keywords:	4408 Political Science; 44 Human Society; Machine Learning and Artificial Intelligence; Generic health relevance
Subjects:	Q Science > QA Mathematics > QA76 Computer software T Technology > T Technology (General) > T58.5 Information Technology
Divisions:	Humanities and Social Science
Publisher:	Cambridge University Press
Date of acceptance:	24 August 2025
Date of first compliant Open Access:	26 March 2026
Date Deposited:	26 Mar 2026 09:23
Last Modified:	26 Mar 2026 09:23
DOI or ID number:	10.1017/cfl.2025.10021
URI:	https://researchonline.ljmu.ac.uk/id/eprint/28297

View Item