The Trump administration is moving to control who gets access to the most advanced frontier models from companies like OpenAI and Anthropic, according to sources familiar with the matter who told CNBC. The White House is taking direct steps to regulate access to cutting-edge AI systems, a shift that rewrites power dynamics between government and the tech industry. The intervention targets frontier models representing the most powerful systems currently in existence. Until now, these companies have largely controlled who gets access to their technology, from enterprise customers to research institutions. The federal government is inserting itself into decisions about who gets to build with the most advanced AI, what applications get greenlit, and which industries get prioritized for access. For OpenAI, already navigating questions about access to GPT-4 and future models, federal control over model access could complicate its partnership with Microsoft. Anthropic, which has positioned itself as the safety-conscious alternative with its Claude models, now faces government gatekeeping on top of its own internal deployment safeguards. The move reflects growing anxiety in Washington about AI's national security implications, as frontier models are systems capable of advanced reasoning, code generation, and potentially capabilities that could be weaponized.