AI Arms Race: The Mutual Assured AI Malfunction (MAIM) Strategy

Rapid AI advancements are reshaping national security. Unstable AI development could disrupt the balance of power and increase the likelihood of great-power conflict, while the proliferation of skilled AI hackers and virologists lowers the threshold for rogue actors to cause catastrophic events. Superintelligence – AI vastly superior to humans in nearly all cognitive tasks – is now anticipated by AI researchers. Just as nations developed nuclear strategies for survival, we now need a coherent superintelligence strategy. The article introduces Mutual Assured AI Malfunction (MAIM): a deterrence regime similar to Mutually Assured Destruction (MAD), where aggressive bids for unilateral AI dominance are met with preventive sabotage. The relative ease of sabotaging a destabilizing AI project—through cyberattacks or physical attacks on data centers—means MAIM already reflects the strategic reality for AI superpowers. Alongside deterrence, states can pursue non-proliferation to prevent rogue actors from acquiring weaponizable AI and bolster their competitiveness through AI-driven economic and military advancements. This three-pronged approach—deterrence, non-proliferation, and competitiveness—offers a robust strategy for navigating the superintelligence era.