kbr-news

Why Claude Fable 5 Stopped: It Was Not a 'Technical Bug,' but a US Government Export Control Order

Anthropic's top-tier AI models, 'Claude Fable 5' and 'Claude Mythos 5,' were blocked for global users just three days after launch due to US government export control directives. Although the core of the guideline was 'blocking foreign access,' Anthropic explained that it was technically difficult to apply this selectively, forcing them to shut down access for all users. While the government took issue with the potential for 'jailbreaking' to bypass model safety guardrails, Anthropic countered that the vulnerability in question was exaggerated and exists across the industry as a whole. This incident demonstrates that frontier AI models are no longer mere technical services, but direct targets of national security and export controls. It also confirms that Korean companies relying excessively on specific AI models or single suppliers may be exposed to service disruption risks resulting from policy shifts.