Trump administration halts Anthropic's Mythos 5 rollout, citing national security risks
the white house forced anthropic to pause its strongest model, mythos 5, restricting access for over 100 us organizations after finding ties to a chinese-linked firm. this signals a new era where federal approval is required for frontier model deployment.
commerce secretary howard lutnick authorized anthropic to restore mythos 5 access to over 100 us cyber defenders and infrastructure providers, determining safeguards were adequate. however, the administration maintained the block on the consumer-facing fable 5 model and kept initial export controls targeting foreign nationals in place. the crackdown was triggered after anthropic shared mythos with a south korean firm suspected of chinese ties, alongside concerns from amazon and the nsa about jailbreaking risks. this intervention coincides with openai delaying its gpt 5.6 models, confirming that federal sign-off is now mandatory for frontier labs.