Как США могут построить устойчивые институты для безопасности пограничного ИИ

Federal oversight of increasingly capable AI systems requires a durable framework that evolves alongside the technology itself. A new strategic blueprint outlines how the United States can move from fragmented state-level approaches to a cohesive national governance model.

Leveraging State Safety Consensus

Governance is already beginning at the state level, creating a foundation for federal action. Current efforts include California’s SB 53, New York’s RAISE Act, and Illinois’s SB 315, which have begun to develop harmonized approaches to frontier AI safety. The proposed national framework seeks to leverage this emerging consensus, integrating these state-level precedents into a broader federal structure alongside the White House’s executive order on Promoting Advanced Artificial Intelligence Innovation and Security.

Strengthening CAISI and National Resilience

Central to this strategy is the strengthening of CAISI as the primary federal institution for managing frontier AI safety. Beyond a single agency, the blueprint calls for a mobilization of broader resilience across the entire government. This cross-agency approach is designed to address the specific national security and public safety challenges posed by the rapid advancement of frontier AI models.

Establishing these institutions now provides the necessary infrastructure to manage the risks of high-capability systems while maintaining the pace of innovation.

Source: A blueprint for democratic governance of frontier AI
Domain: openai.com