The study of how to ensure safe and fair behavior when multiple AI agents interact, compete, or cooperate in shared environments.