Anthropic globally suspends access to its flagship Claude Fable 5 model following a U.S. Commerce Department export control order over jailbreak vulnerabilities.
Anthropic has suspended global access to its next-generation language model, Claude Fable 5, following an export control directive from the U.S. Department of Commerce. The suspension was triggered by potential jailbreak vulnerabilities discovered in the model. Claude Fable 5 is Anthropic's flagship frontier model, which was released as the safe public version of its restricted "Mythos-class" architecture.
Anthropic's forced suspension of its most capable model shows that the U.S. government is actively enforcing export controls on frontier AI capabilities based on safety and vulnerability profiles.
- –This marks a major moment where a civilian LLM is suspended globally due to U.S. government intervention over safety and jailbreak issues.
- –Claude Fable 5 shares core weights with Mythos, a restricted government/research model; jailbreaking Fable 5 could leak Mythos-class capabilities to unauthorized entities.
- –The incident highlights the extreme difficulty of wrapping highly capable frontier models in secure guardrails without sacrificing their advanced agentic functionality.
DISCOVERED
2h ago
2026-06-13
PUBLISHED
2h ago
2026-06-13
RELEVANCE
AUTHOR
Prompt Engineering