Anthropic disabled access to its newest frontier models—Fable 5 and Mythos 5—after the U.S. Commerce Department imposed national security export controls that barred distribution to foreign nationals. The directive extended beyond users outside the U.S. to any foreign national located in the U.S., including some of Anthropic’s own employees. Anthropic said the company was told the action followed government findings about a technique to bypass safeguards, designed to prevent powerful cybersecurity capabilities. Anthropic argued the cited “jailbreak” was narrow defense-oriented prompting and said the compliance standard could effectively halt frontier model deployments. The move comes amid broader questions about how governments apply safety rationales through export controls, and it raises operational concerns for globally staffed AI providers and universities testing AI systems across international user populations. For colleges and research institutions, the development affects access to model capabilities for international students, visiting researchers, and cross-border collaborations, with potential knock-on effects for coursework, lab research, and AI-enabled services.