The Trump administration moved to restrict access to Anthropic’s newest frontier models after cybersecurity concerns reportedly surfaced during testing tied to national-security export controls. Commerce Department action barred Anthropic from distributing Fable 5 and its underlying Mythos 5 to foreign nationals, including people working inside the U.S. who are not U.S. citizens. Anthropic said it pulled the models offline to comply and disputed characterizations that the research amounted to a full jailbreak. Amazon CEO Andy Jassy and Amazon researchers were reported as key actors in raising the issue after prompts revealed restricted information about cyberattacks. Multiple outlets reported that government officials asked Amazon for feedback on Anthropic’s models, while Anthropic said it does not allow access from within China and that its safeguard bypasses were narrow. The incident has quickly become a reference point for how governments may treat AI distribution as both cybersecurity and export-control policy. Canadian Prime Minister Mark Carney warned the episode highlighted the risk of overreliance on a small number of American AI providers, arguing that diversification is needed as states respond with increasingly forceful national security restrictions. The episode also triggered debate among technologists and policymakers about whether export-control standards should be clearer, more transparent, and applied consistently across the industry.