US government actions have focused on temporary export controls and pre-release reviews for closed frontier models from labs like Anthropic and OpenAI, while explicitly exempting or promoting open-weight models in recent executive orders and the Open-Source AI Leadership Act. These policies aim to bolster US competitiveness against Chinese open models and support developer ecosystems without restricting domestic releases. No open-source bans or mandatory licensing for weights have advanced, reflecting trader consensus that broad restrictions remain unlikely before year-end. Potential swing factors include a major verified misuse incident or new legislation targeting adversary models, though current frameworks prioritize voluntary collaboration and carve-outs for open releases.