The U.S. government has issued an export control directive requiring Anthropic to disable access to its powerful AI models Fable 5 and Mythos 5 for all foreign nationals, citing national security risks embedded in frontier artificial intelligence systems.

The directive, issued Friday, marks the most aggressive use of export restrictions against a commercial AI company to date. Anthropic confirmed the order in a brief statement, saying it is complying with the suspension while the company reviews its options.

The two models are among the most capable in the industry, ranking near the top of independent benchmarks for reasoning and coding tasks. Their sudden cutoff affects thousands of researchers, universities, and businesses outside the United States that had licensed the software through Anthropic's enterprise channels.

National security officials have grown increasingly concerned that frontier AI models could accelerate foreign military and intelligence programs. The new restrictions echo earlier export controls on advanced semiconductors, but they place the U.S. government directly in the role of qualifying which foreign users may access commercial AI services.

Legal scholars note that the move raises novel questions about the reach of export law into software distribution and cloud services. Anthropic has not indicated whether it intends to challenge the directive in court, though the company has previously taken legal action against the Pentagon over procurement disputes.

The policy shift arrives as the race for AI dominance intensifies. European regulators are drafting their own AI governance framework, and Chinese state-backed labs continue to expand access to comparable models. The suspension could accelerate the fragmentation of the global AI ecosystem into distinct, jurisdiction-bound systems.

Image source: v3b.fal.media