TechCrunch argues three decades of failed export controls on security software cast doubt on any attempt to block Anthropic’s Mythos model.