On June 12 at 5:21pm ET, the US Government issued an export control directive requiring Anthropic to immediately suspend all access to Claude Fable 5 and Mythos 5 for all users worldwide — including foreign-national employees. The government cited awareness of a "potential narrow, non-universal jailbreak": specifically, asking the model to read code and identify flaws. The suspension affects claude.ai, the Claude API, Claude Code, and Claude Cowork; all other Anthropic models (Opus 4.8, Haiku 4.5, etc.) remain fully operational.
Anthropic issued a public statement confirming compliance while explicitly disagreeing with the decision. The company notes that the alleged jailbreak method is "widely available from other models including OpenAI's GPT-5.5" and that Fable underwent thousands of hours of red-team testing with safeguards described as "substantially more effective than those of any previously deployed model." Anthropic's key objection: if this standard were applied industry-wide, it "would essentially halt all new model deployments for all frontier model providers." Simon Willison independently observed he still had Fable access hours after the directive was issued, suggesting a rolling enforcement rollout.
model: "claude-opus-4-8"), and implement a model-version health check at pipeline startup so future suspensions are detected automatically within minutes rather than hours.