AI increases likelihood of Cyber-attacks and Cyber war

The global proliferation of artificial intelligence (AI) technology has shortened the distance between cyber warfare and military conflict. Modern warfighters have developed and embraced autonomous weapons programmes to fight wars across domains. As a result, AI technologies have raised the chance that military personnel may use autonomous systems to launch cyber operations during military operations (Guembe ). AI programmes may also enable military officials to deploy cyber means more effectively to destroy their adversaries and consequently, deepen human reliance on the use of cyberweapons as a means of planning and winning battles. The weaponisation of AI technologies, ranging from image recognition to text analysis to game playing, leaves tremendous implications for the use of violence. Yet we do not know how exactly AI, cyberwarfare, and traditional military conflict are connected.

This question is critical because there is no scholarly effort to link two important phenomena taking place – AI reshaping strategic discourse and “cross-domain operations.” In the literature of international security, there is little to no work on how AI would affect war fought across cyber and military domains. As a result, we do not know how machines may affect the balance of power between major powers like Russia, the United States, and China that have increased investment into AI research and development (R&D) and intensified the AI arms race (Kostyuk and Gartzke ). In this paper, I take up the notion of “cross-domain warfare” to explore how AI affects the relationship between cyber and military operations that states carry out.

My research points to two major observations. First, in the absence of AI programmes specifically designed to help humans fight cross-domain war, there is no “straight line” we can draw between AI and cross-domain war. Research shows that cross-domain war has been rarely fought relative to a large number of cyber and military operations held separately (Katagiri ), and there is no evidence that humans would be able to use AI to reverse the trend. When AI shapes human conduct of cyber operations and military operations, it would do so in each of the domains separately. If any, the way AI would shape the chance of cross-domain war is neither causal nor correlated because internal flaws in many existing systems are prone to generate an excess amount of unpredictability – and what Arthur Holland calls “known unknowns” – to establish any kind of linearity in war outcomes (Holland ). In fact, it is the very uncertainty about AI’s capabilities and limitations that poses the greatest influence on the way human warriors fight cross-domain war.

Second, while AI alone is unlikely to become a major cause of cross-domain war in the near future, it is likely to affect the chance of cross-domain war in two ways. In this article, I will hypothesise that first, AI is likely to cause peacetime change in the balance of cyber and military power in two ways; they reshape the distribution of military capability between nations, which in turn affect the likelihood of cyberwar, and upset the balance between defence and offence in cyberspace. While the peacetime redistribution of capability does not necessarily lead to cross-domain conflicts, it allows countries to prepare for it. Second, limitations with existing AI programmes are likely to increase the chance of inadvertent conflict. This is not because state actors will heed AI recommendations on the use of force but because AI’s programmatical flaws, such as bias towards minorities and confounding problems, will drive state actors into disagreement even when they do not wish to. The implications of this analysis are that, while AI systems are unlikely to affect the wartime performance of nations, they are likely to affect the “peacetime” redistribution of capability in the cyber and military domains and the chance of such conflict in the future.

This article contributes to the literature of AI and cross-domain warfare by linking the two fields of inquiry. It does so by exploring how AI may reshape the balance of cyber and military power and reinforce a bridge between the two domains, represented by works done by scholars like James Johnson () and Martin Libicki (, 199–201). The article also seeks to bolster our understanding of how AI may affect traditional interstate rivalry and increase the likelihood of cyber escalation and accidental warfare, areas of literature developed by Kenneth Payne, Erica Lonergan, Shawn Lonergan, and RAND researchers (Lonergan and Lonergan ; Morgan ; Payne ). This article makes these contributions by using a qualitative methodology based on the use of recent literature and by presenting hypotheses about how AI may affect the balance of power and the chance of unintended escalation between nations. The article gathers expert views on the hypotheses to advance our knowledge of how AI affects international conflict in physical and digital domains. It does not empirically test these hypotheses but treat some of the hypotheses in the domain of speculation. With the limited purpose and scope, the article seeks to advance the state of our knowledge about how autonomous technologies may become more relevant in the future by pushing up the likelihood of hostility. This analysis is organised in four sections. First, I will set the scope of analysis by defining the contour of AI’s flaws, limits, and capabilities. Second, I examine the state of our knowledge about cross-domain warfare and how AI affects it. Third, I elaborate on the ways by which AI is likely to shape cross-domain warfare. Finally, I offer my conclusion.