During 2024, efforts to address the governance of military artificial intelligence (AI) have gained momentum. Yet in the same year, we have also witnessed the growing use of AI decision support systems during armed conflict, and it is becoming clearer that such systems may pose a significant challenge to peace and stability. These developments raise questions about the current approach toward military AI governance.

In this post, Elke Schwarz, Professor of Political Theory at Queen Mary University London, argues that efforts toward governance are complicated by a number of factors intrinsic to contemporary AI systems in targeting decisions. She highlights three in particular: (1) the character of current AI systems, which rests on iteration and impermanence; (2) the dominance of private sector producers in the sector and the financial ethos that grows from this; and (3) the expansive drive implicit in AI systems themselves, especially predictive AI systems in targeting decisions. These realities of AI suggest that the risks are perhaps greater than often acknowledged.