Trump administration restricts OpenAI’s GPT5.6 Sol and Anthropic’s Mythos 5 to approved customers under security review

The Trump administration has ordered OpenAI and Anthropic to roll out their latest AI models, GPT-5.6 Sol and Mythos 5, on a restricted basis, limiting access to a small set of customers approved by the administration over cybersecurity concerns. The move follows an AI oversight executive order signed in June that set a voluntary 30-day security review framework ahead of public releases. Anthropic was previously temporarily blocked over alleged weaponization risks tied to Mythos, and its Fable 5 model was also taken offline for two weeks. The restrictions have directly affected both companies’ IPO plans and added to market concerns about commercialization timelines and regulatory uncertainty.