In a nutshell: Anthropic was ordered on national security grounds to immediately disable Fable 5 and Mythos 5 after the government was presented with a jailbreak technique that Anthropic classifies as widely prevalent among competitor models as well.
The US government has instructed Anthropic to disable the Fable 5 and Mythos 5 models for all users worldwide — including employees with foreign nationality. The reason is a jailbreak method that the government considers a security risk.
The US government issued an export control directive to Anthropic on June 13, 2026 at 17:21 ET, suspending access to Fable 5 and Mythos 5 for all users — regardless of nationality or geographic location. The company subsequently disabled both models for all customers. Other Anthropic models remain available.
The government’s justification refers to national security. According to Anthropic’s statement, it concerns a potential jailbreak in which the model is prompted to analyze a specific code repository and remediate software vulnerabilities. The company has examined this technique and classifies the security gaps identified by it as known and straightforward in structure. Multiple publicly available models, including OpenAI’s GPT-5.5, can identify the same vulnerabilities without specialized jailbreak methods.
For Chief Data Officers and security officers, the directive has an immediate operational impact: dependencies on Fable 5 or Mythos 5 must be migrated to other models. Anthropic announced it will publish further details on the situation in the coming 24 hours. The measure demonstrates increasing regulatory oversight of AI model deployment and raises questions about the distinction between technical vulnerabilities and export control risks.
Source: simonwillison.net · Published June 13, 2026
Lumi AI News — AI-assisted curation pursuant to Art. 50 EU AI Act. Paraphrasing and classification by Lumi News Pipeline v1.7.1.