What happened
On June 12, 2026 at 5:21 PM ET, Commerce Secretary Howard Lutnick sent a letter to Anthropic CEO Dario Amodei invoking national security authorities to issue an export control directive suspending all access to Claude Fable 5 and Claude Mythos 5 by any foreign national — whether inside or outside the United States, including Anthropic's own foreign-national employees. Because Anthropic cannot filter foreign from domestic users in real time, it abruptly disabled both models for all customers globally. Access to all other Anthropic models (e.g. Claude Opus 4.8) was unaffected. The administration acted after another company claimed it had jailbroken Mythos 5, alarming officials about potential exploitation of the model's exceptional cybersecurity and vulnerability-finding capabilities. Anthropic publicly disputed the rationale, saying the demonstrated technique was narrow and that comparable capabilities exist in other publicly available models such as GPT-5.5. This is the first time the US government has applied export controls directly to a commercial AI model (as opposed to chips or hardware).
Why it matters
This is a landmark Tier A enforcement action — the first-ever application of US export controls to a specific AI model. It establishes that frontier AI models can be treated as controlled commodities akin to munitions, subject to immediate revocation of access with no advance notice. Every enterprise relying on frontier AI models now faces the operational risk that government action can disable mission-critical AI access overnight. It also signals that jailbreak disclosures — even narrow ones — can trigger federal intervention, creating new incentives around responsible disclosure and safety testing regimes. The precedent could extend to other labs (OpenAI, Google DeepMind) and other models.
Action needed
Enterprises using Anthropic Fable 5 or Mythos 5 must immediately switch to alternative models (e.g. Claude Opus 4.8 or competing providers). All organisations relying on frontier AI should develop multi-model contingency plans and assess geopolitical/export-control risk in AI procurement. AI legal and compliance teams should monitor Commerce/BIS for licence requirements if seeking to restore access to Fable 5/Mythos 5.