To Steer the AI Frontier, Washington Must Build Its Testing Power

Current AI policy debates often rely on speculation, creating a need for evidence-based governance. The US government's ability to evaluate frontier AI models has not kept pace with the companies building them, creating a critical gap. To regulate effectively, Washington must develop its own independent testing capacity and not rely on AI labs to grade their own homework on safety and security. This requires a clear mandate, pre-deployment evaluation access, and significantly increased funding for public-private testing to identify real-world risks like cyber vulnerabilities.