Anthropic Blocks Fable 5 and Mythos 5 for National Security

Anthropic Cuts Access to Fable 5 and Mythos 5 for Security Reasons

On Friday evening, Anthropic received an order from the U.S. government to block access to its artificial intelligence models, Fable 5 and Mythos 5, for all foreign nationals. This measure affects both those residing in the United States and abroad, and also includes the company's employees. This decision was motivated by national security concerns.

To comply with this directive, Anthropic has decided to completely cut off access to these models for all of its clients. However, the company expressed its dismay at the lack of details provided by the government regarding the specific threats to national security.

Minor Vulnerabilities Mentioned

In a statement, Anthropic clarified that the only information received about potential vulnerabilities in its models was communicated verbally. These flaws, described as minor, are also present in other AI models, such as GPT 5.5. The company emphasized that the identified vulnerabilities did not pose a significant danger and were accessible through other systems.

Anthropic also detailed the measures implemented to secure Fable and Mythos. The company is working in collaboration with the U.S. and U.K. governments and has modified its data retention policy to better track attempts to misuse its models.

No Concerning Jailbreak Disclosure

Anthropic stated that it has not received any disclosures regarding a potential non-universal jailbreak that could have led to harmful consequences. Reported incidents turned out to be harmless responses or minor discoveries with no particular advantage for Mythos.

This situation highlights the challenges faced by tech companies regarding security and compliance with government requirements, while also underscoring the importance of transparency and communication in managing risks associated with artificial intelligence.