Anthropic has suspended access to its advanced AI models Fable 5 and Mythos 5 for all customers globally following a government order citing national security risks, marking a rare move to restrict AI model availability across borders.
- Government order blocks foreign access to Anthropic's Fable 5 and Mythos 5
- Anthropic reports minor vulnerabilities and no significant harmful jailbreaks
- Company enhanced data policies and cooperated with US and UK governments pre-order
What happened
On a recent Friday evening, the US government formally instructed Anthropic to restrict access to its flagship AI models, Fable 5 and Mythos 5, from all foreign nationals regardless of location. This measure effectively cut off international users and Anthropic employees outside the US from utilizing these tools. In response, Anthropic complied by fully disabling access to both models for all customers globally.
The directive arose from alleged national security concerns, although Anthropic noted it had not been provided with a clear or detailed rationale from authorities. The company stated that the government only verbally communicated possible security risks linked to potential model jailbreaks, which Anthropic described as minor and not unique to their systems.
Why it matters
This unprecedented enforcement highlights growing governmental scrutiny of advanced AI capabilities, especially regarding their distribution and use by foreign entities. It illustrates the tension between innovation openness and national security considerations, particularly as AI models become more powerful and widely accessible.
Anthropic’s public remarks reveal a disconnect between regulatory demands and transparency, emphasizing the difficulties AI developers face when balancing compliance with user continuity and trust. The move could set a precedent for future restrictions on AI technology access based on geopolitical or security grounds.
What to watch next
Stakeholders will be closely monitoring how Anthropic and other AI companies navigate such regulatory orders going forward and the potential for broader government-driven limitations on AI model availability internationally. Attention will also focus on any further disclosures or evidence that clarify the specific national security threats cited by authorities.
Observers should track updates on policy adjustments, including Anthropic’s data retention changes aimed at detecting malicious use attempts, and whether this incident catalyzes industry-wide shifts in AI governance or increased cooperation between governments and AI developers across borders.