Exploring the Security Implications of Artificial Intelligence in Military Contexts

Artificial Intelligence (AI) has been described as “revolutionary” for modern warfare. This thesis focuses on the complex and evolving military AI innovation landscape underexplored to date. Existing research on the implications of AI in military contexts is emerging yet nascent despite the apparent pivot to drive the adoption and use of AI-enabled systems in militaries across the globe. This thesis aims to identify the security implications of military AI innovation and explore U.K., U.S., and NATO approaches to adopting AI in a military context. This research employs a multiple methods approach to engage with this rapidly emerging field, analysing available non-classified literature and policy documentation, observational methods, and expert interviews. This thesis finds that defence-focused communities are increasingly interested in military AI innovation and associated security implications, recognising that AI will have a far-reaching impact beyond creating new capabilities. This research highlights a growing awareness of AI among security-focused defence practitioners and policy experts that states must adopt AI to secure or maintain perceived military advantages against adversaries. There are significant challenges relating to AI in military contexts, which include: (1) technical aspects such as cyber security and data protection, (2) organisational aspects including barriers to procurement and the culture of military innovation (alongside wider knowledge and awareness), and (3) strategic aspects including the increased speed of warfare in which humans may not have time to understand or interfere with AI-enabled processes, and potential widening in capabilities in light of “AI arms race” activity. Finally, this research highlights the role of international military organisations as a productive mechanism for developing norms and common approaches to the use of military organisations and explores NATO’s opportunities and challenges to contribute to this domain. These discussions form methodological and policy-related contributions to scholarship and aim to inform academic and policy-focused communities. This research will be of particular interest to defence professionals dedicating their efforts to mitigate the negative security implications of AI in military contexts.