Anthropic has previously emphasized AI safety and national security risks, and current concerns regarding distillation of Chinese models continue its consistent closed-source and regulation-friendly approach. U.S. policymakers are mobilizing resources towards domestic closed models through export controls and investment reviews, with the strategic motive of preventing key AI technological capabilities from rapidly diffusing to competitors. Similar to the impact of semiconductor export controls on Huawei, the current open weight AI sector is undergoing a transformation phase amid accelerated U.S.-China tech decoupling, with policy barriers reshaping the global model training and distribution landscape. This represents a change in regulation: shifting from open innovation to prioritizing national security, restricting knowledge transfer paths through distillation, forcing open weight models to find a balance between compliance and innovation, and accelerating the concentration of industrial capital towards protected closed-source players. During the policy testing period, the survivors are not the most open, but those who adapt best to the rules.