Strategic Preemption Under Shared Catastrophic Risk: The Suicide Region and the Race to Artificial General Intelligence

Abstract

We analyze a continuous-time preemption game with shared catastrophic externalities. When the cost of catastrophe is embedded in both players' payoffs, the risk term cancels out in the equilibrium indifference condition. This creates a "suicide region" where competitive pressures force rational agents to deploy despite negative risk-adjusted net present values. We apply this framework to the race for artificial general intelligence (AGI). We show that this suicide region widens as the cost of systemic ruin grows: higher catastrophic risk does not deter the race but instead enlarges the set of conditions under which rational actors deploy despite negative social value. We characterize the resulting welfare distortion against a social planner's benchmark and demonstrate how two complementary mechanisms - private liability and prize-sharing - can close the suicide region. Private liability raises the cost of unsafe deployment while prize-sharing reduces the strategic imperative to deploy first. "Warning shots" (sub-existential disasters) will fail to deter AGI acceleration, as the winner-takes-all nature of the race remains intact.