On July 18th, CNBC broke the news that the Trump administration has rolled out fresh measures to ramp up its grip on the AI model release process. These measures involve designating specific companies and entities that are allowed to access the latest state-of-the-art models. Previously, this decision-making authority was in the hands of U.S. AI behemoths. Anthropic and OpenAI, for instance, had the autonomy to decide who could use their models. Typically, their user base included large enterprise clients. However, the U.S. government has recently started to tighten the regulatory screws. Anthropic, for one, has chosen to release its most potent Mythos cybersecurity model to only a select group of partners through its 'Glasswing Project.' OpenAI, too, has followed suit, imposing access restrictions on GPT-5.6 in line with government requests. Despite White House officials asserting that the government won't greenlight AI model releases by private companies and that the timing and scope of releases are up to the companies themselves, the Trump administration did just that last month. It blocked access to Claude Mythos 5 and Fable 5, citing 'national security concerns.' It was only after weeks of intense negotiations that access was restored. Moreover, to toe the government line, OpenAI has now limited the availability of its new AI models to 'trusted partners' only.