Anthropic shut down access to its Fable 5 and Mythos 5 models Friday night, just days after their launch. The abrupt removal followed a US Commerce Department export control directive restricting their use outside the United States. Anthropic stated disabling the models was the only way to ensure immediate compliance with the government order.

Administration officials raised concerns about a jailbreak that bypasses safeguards designed to block prompts on cybersecurity, chemistry, and biology topics. The directive provides time for national security systems to strengthen protections against such threats. Officials suggested this hardening could complete within weeks.

Anthropic's announcement referenced only "verbal evidence" of a narrow jailbreak involving code review for software flaws. The company characterized vulnerabilities found as "minor" and "relatively simple," noting other publicly available models like GPT-5.5 have similar capabilities. Other Anthropic models remain unaffected by this action.

This shutdown underscores mounting pressure on AI companies to balance rapid deployment with national security requirements. The incident may establish precedent for how emerging AI models navigate government oversight while maintaining competitive advantage in global markets.