US export controls on Anthropic's Mythos 5 and Fable 5 models were lifted, with Mythos 5 now available to vetted US organizations. However, strict guardrails on frontier AI models from Anthropic and OpenAI are hindering legitimate offensive cybersecurity research. This forces network defenders and security researchers to increasingly rely on less restricted, often foreign-owned, open-source AI models for vulnerability discovery and exploit development.