The Center for AI Safety Action Fund (CAIS AF) submitted recommendations on U.S. AI policy in response to the administration’s Request For Information on the Development of an Artificial Intelligence (AI) Action Plan.
“The stakes for AI policy could not be higher. The U.S. has a narrow window to shape global norms, secure critical technologies, and stay ahead in the race to build a world-changing technology,” said CAIS AF. “The AI Action Plan will guide U.S. tech and national security policy for years to come, and we’re grateful for the opportunity to offer policies that protect U.S. leadership and global stability.”
Our response centers on three interconnected pillars based on the recently published Superintelligence Strategy—authored by CAIS founder Dan Hendrycks, former Google CEO Eric Schmidt, and ScaleAI CEO Alexandr Wang—that we believe must form the foundation of any robust AI policy: Nonproliferation, Deterrence, and Competitiveness.
- Nonproliferation: The U.S. must secure the AI chip supply chain through a comprehensive approach including robust export controls, mandatory location verification technologies for advanced AI chips, and strengthened enforcement to prevent these critical computational resources from falling into the wrong hands, such as rogue actors. This includes implementing a comprehensive licensing and notification regime for export-controlled AI chips and potentially repositioning the AI Safety Institute to focus on export control enforcement.
- Deterrence: The U.S. and its allies need to deter strategic competitors from pursuing destabilizing AI projects. This requires strengthening intelligence collection on foreign frontier AI capabilities and developing offensive cyber capabilities to credibly deter rivals from high-risk AI development programs. We also recommend stronger reporting requirements for developers of frontier AI models to enhance government visibility into emerging capabilities and potential risks.
- Competitiveness: To maintain our edge in the AI race, the U.S. must focus on strengthening its AI competitiveness. This includes accelerating domestic AI chip production, implementing streamlined immigration pathways for top AI talent, and securing guaranteed drone and robotics supply chains. Competitiveness is a national security imperative because economic security is national security
With these recommendations, the United States can protect our economic and national security and enhance our global AI leadership.
