Exploring Instability Risks in the U.S.-China AI Rivalry

The race to develop ever-more-advanced artificial intelligence (AI) is becoming increasingly prominent in the ongoing strategic rivalry between the United States and China. This intensifying technological competition raises the possibility that one or both of the contestants might seek to improve its prospects in the race — either to secure a lead or to keep the other from achieving artificial general intelligence (AGI) first — by attacking elements of its competitor's AI ecosystem using military force, cyber warfare, or other means beyond the bounds of normal peacetime statecraft. Anticipating and managing these risks is important for both technology and national security policymakers, most of all because of the potential for such preventive actions to lead to armed conflict and potentially catastrophic escalation.

As part of RAND's ongoing research into the geopolitical implications of AGI, a research team has developed and begun playing a game called Breakwater. Breakwater is designed to shed light on several interrelated questions: Under what conditions might competitors in the AI race resort to preventive actions, including cyberattacks or military force, to slow a rival's progress? How do beliefs about the trajectory and timeline of AI development shape strategic behavior and risk tolerance? What factors tend to stabilize or destabilize the competition as it unfolds?

This paper provides a brief overview of the game design and a summary of initial observations and hypotheses from early plays of the game.