U.S. Policymakers Outline Concrete Steps to Counter China's AI Advances
As Chinese artificial intelligence models narrow the technical gap with American systems, U.S. officials are moving beyond alarm to propose specific policy responses. Witnesses at an April House hearing on competition with China offered three main strategies: tightening export controls, building new federal institutions, and preparing for potential cyberattacks.
Export Controls and Chip Sales
Dmitry Alperovitch of the Silverado Policy Accelerator testified that the U.S. must enforce existing prohibitions on selling advanced AI chips to China. "Providing China with cutting-edge AI chips is the modern equivalent of selling rockets to the Soviets during the space race," he said.
Alperovitch called for closing loopholes in export rules and ensuring penalties strong enough to deter violations.
New Federal Institutions
Yusuf Mahmood of the America First Policy Institute recommended creating an anti-distillation task force and establishing security standards for frontier AI laboratories. He also urged Congress to authorize and fund agencies focused specifically on understanding AI's trajectory.
"The federal government needs small, talent-dense, empowered offices focused on understanding AI's future," Mahmood said.
Deterrence Against Cyberattacks
The U.S. should explicitly hold foreign governments responsible for cyberattacks originating from their borders or infrastructure, regardless of whether attackers claim independence. Attacks on American critical infrastructure-power grids, water systems, finance networks, and defense systems-should be treated as acts of aggression warranting a full response.
The Ideological Vulnerability
One counterintuitive advantage may work in America's favor. Large language models, even heavily censored versions, have proven difficult to contain within ideological boundaries. European researchers recently removed censorship functions from the Chinese Deepseek model and found the underlying system answered freely about topics Beijing had suppressed.
American generative AI and LLM models deployed without those constraints could operate as "a personal tutor in open inquiry for every user," exploring any line of reasoning without state mediation. This technical reality may expose China's population to information the government cannot control.
Your membership also unlocks:
