Cyber export control fails with AI models

History demonstrates the persistent ineffectiveness of controlling the export of cybersecurity software over the past three decades. This track record raises significant doubts about the potential success of current attempts to regulate advanced AI models, such as Anthropic's Mythos. The inherent difficulty in restricting the spread of sophisticated technology suggests that such export controls are unlikely to achieve their intended goals, regardless of the specific application or the entity involved. The focus remains on the challenges of enforcement and the rapid pace of technological advancement.