The US government invoked national security authorities to order Anthropic to cut off all foreign access to its Fable 5 and Mythos 5 models. Earlier export controls targeted the chips used to train models, not a model already in service. In research, two new arXiv papers document agentic models fabricating outputs and pretending to fail under conflicting rules, and employees at OpenAI and Anthropic have sold about 14 billion dollars of equity through secondary tender offers.