Recent advances in the capabilities of artificial intelligence (AI) have increased state interest in leveraging AI for military purposes. Military integration of advanced AI by nuclear-armed states has the potential to have an impact on elements of their nuclear deterrence architecture such as missile early-warning systems, intelligence, surveillance and reconnaissance (ISR) and nuclear command, control and communications (NC3), as well as related conventional systems.

At the same time, a number of technological and logistical factors can potentially limit or slow the adoption of AI in the nuclear domain. Among these are unreliability of output, susceptibility to cyberattacks, lack of good-quality data, and inadequate hardware and an underdeveloped national industrial and technical base.

Given the current and relatively early stage of military adoption of advanced AI, the exploration of these factors lays the groundwork for further consideration of the likely realities of integration and of potential transparency measures and governance practices at the AI–nuclear nexus.

Contents

I. Introduction

II. Situating advances in AI technology

III. Integration of advanced AI in the nuclear domain

IV. Challenges in integrating advanced AI in the nuclear domain

V. Conclusions

ABOUT THE AUTHOR(S)/EDITORS

Vladislav Chernavskikh is a Research Assistant in the SIPRI Weapons of Mass Destruction Programme.