The U.S. government moved to restrict foreign access to Anthropic’s frontier AI models Fable 5 and Mythos 5 after cybersecurity researchers identified a jailbreak path tied to guardrails. Anthropic disabled the models for all users subject to deemed-export rules, triggering an international scramble over “sovereign AI” and creating disruption for researchers and organizations that relied on access. Europe’s policymakers warned the decision functions like a “kill switch,” renewing calls for more independent model, computing, and data capacity. Meanwhile, Anthropic’s efforts to reverse the controls included high-level engagement in Washington, but no compromise has been reached. The fallout is spilling into higher-education tech infrastructure planning: universities and research labs face uncertainty over tool availability, cross-border collaboration, and compliance obligations when model access is tied to nationality and export licensing.