Abstract
This article argues that the deployment of artificial intelligence, even in its strong fully autonomous form, does not have significant consequences for the application of the jus ad bellum framework, and in particular the armed attack threshold. As with all new weapons technologies, the deployment of artificial intelligence in the resort to force requires consideration of what “imminence” means in this context. However, artificial intelligence is fundamentally a means of making decisions to resort to force or contributing to the reasons for which such decisions are made by humans. Unlike the jus in bello, the jus ad bellum is not concerned with how or why decisions about resort to force are made, and the definition of “armed attack” contains no “subjective” element, such as fault or intention. Whether a use of force is initiated by human or artificial intelligence, and whether either of those intelligences is mistaken, is therefore not something that can affect whether a use of force meets the armed attack threshold. The mere use of artificial intelligence also seems incapable, by its very nature, of affecting the gravity of a use of force, understood as its scale and effects, such as to make it more or less likely to reach the armed attack threshold. Many implications of the deployment of artificial intelligence for the jus ad bellum will be determined by other rules of general international law, notably the customary international law rules on attribution. However, it would not be appropriate for the jus ad bellum to pre-empt such a development by modifying its own standards.
html
