The United States and its allies must secure a lead in AI development over authoritarian governments like the Chinese Communist Party (CCP) to prevent unprecedented repression and shifts in global power. Access to advanced computer chips ("compute") is paramount, where American companies currently hold an advantage, maintained by export controls.
Chinese AI labs, however, exploit loopholes and conduct distillation attacks to mimic US innovations, remaining competitive. The article outlines two 2028 scenarios: one where America tightens export controls, disrupts attacks, and accelerates democratic AI adoption, ensuring democracies set global AI norms and enable safety engagement; the other where inaction allows China to catch up or surpass, leading to authoritarian regimes shaping AI rules and enabling automated repression. Urgent policy action is required to lock in a 12-24 month lead in frontier capabilities by 2028, crucial for national security and responsible AI governance. A close race risks disincentivizing safety measures, potentially leading to rapid, unguarded deployment of dual-use models.
No comments:
Post a Comment