“Under the "Mutually Assured AI Malfunction" (MAME) framework, nations like the U.S., China, and Russia would deter each other from developing destabilizing "superweapon" AIs by maintaining a credible threat of a preemptive cyberattack on rival data centers.”