Exploring the Constraints on Artificial General Intelligence: A Game-Theoretic No-Go Theorem