U.S. Forces Anthropic Models Offline
Coverage from NBC News, Inside AI, and others

The U.
S. Commerce Department reportedly ordered Anthropic to block foreign-national access to its Fable 5 and Mythos 5 models after officials raised national-security concerns about a jailbreak and the models’ potential cyber capabilities. Anthropic disabled both models for all customers to comply, disrupting enterprise access and exposing tensions between frontier-model safety, export controls, international staffing, and government oversight. The company disputes the severity of the reported vulnerability and argues that the action lacked a transparent, technically grounded process.
The story is largely stable, but the framing has sharpened around Commerce Department enforcement and the possibility of licensing and penalties. The current version also more explicitly disputes the scope of the vulnerability and emphasizes the broader implications for frontier-model regulation.
The story has shifted from a narrower access dispute to a clearer account of binding U.S. export-control enforcement, with Anthropic now saying it must contest the government’s opaque process and technical basis. The new version also adds Amazon as a reported participant raising concerns and emphasizes broader commercial and workforce disruption.
