TITLE: US Government Blocks Foreign Access to Anthropic's Most Powerful AI DATE: 2026-06-13 COMPANY: Anthropic TOPIC: AI Security SUMMARY: Commerce Secretary Howard Lutnick sent a letter to Anthropic CEO Dario Amodei on June 12, 2026, placing Fable 5 and Mythos 5 under US export controls that restrict access to US persons only. The action was triggered by a third party claiming to have jailbroken the Mythos model, prompting national security concerns in the Trump administration. Both models were released to the public just three days earlier on June 9. WHAT CHANGED: Claude Fable 5 and Mythos 5 were announced on June 9, 2026 as Anthropic's most capable models to date. Fable 5 was made publicly available and described by Anthropic as a version of Mythos with additional safety measures. Mythos 5 itself remained in controlled preview through Project Glasswing, a programme giving selected critical infrastructure organisations access to the more powerful underlying model. Three days after launch, on June 12, Commerce Secretary Howard Lutnick sent a letter to Dario Amodei formally placing both models under US export controls. According to reporting by Axios, an administration official confirmed the action followed a report from an unnamed company claiming it had successfully jailbroken the Mythos model. The administration said this raised concerns about national security risks if the models remained accessible outside US jurisdiction. The Commerce Department had reportedly contacted Anthropic before the launch and requested a delay. Anthropic proceeded with the release on June 9 as planned. The export control letter followed 72 hours later. The controls are structured under existing US export law. They restrict access to Fable 5 and Mythos 5 for any person outside the United States and for foreign nationals accessing the models from within the country. The practical enforcement mechanism, and how Anthropic plans to implement it technically, had not been publicly detailed as of this writing. WHY IT MATTERS: Access to frontier AI is now a geopolitical variable. Until June 12, most organisations assumed that access to cutting-edge AI models was a commercial question: could you afford the subscription or API costs? That assumption no longer holds for the most capable models. If you are based outside the US or employ non-US staff, access to Fable 5 is now a policy question, not a commercial one. International businesses face immediate compliance risk. Any organisation using Fable 5 with team members accessing it from outside the United States, or with staff who are foreign nationals, may already be in breach of the controls. The scope of "foreign persons within the country" is broad and can cover visa holders, contractors, and permanent residents who are not US citizens. The trigger reveals the real concern. The stated cause was a jailbreak of the Mythos model. This signals that the US government now views frontier AI capability as a national security asset, comparable to advanced semiconductors or encryption technology. Once that classification is applied, the regulatory trajectory becomes predictable: more controls, not fewer. This sets a precedent for the entire industry. No specific AI model has previously been named in US export controls. Every AI lab building at the frontier now faces the prospect of similar action. Businesses choosing AI vendors should factor regulatory exposure into their procurement decisions, particularly if they operate in multiple countries. Model selection strategy must account for access continuity. If your most critical workflows depend on Fable 5 and your team includes non-US staff or international operations, you now have a single point of failure in your AI stack. Operational resilience requires a tested fallback that does not depend on geopolitical stability. The speed of regulatory action will accelerate. Fable 5 launched on June 9. Export controls arrived June 12. That is a 72-hour window between model release and government restriction. Future frontier model launches may carry access uncertainty from day one. DAVID & GOLIATH ANALYSIS: The most significant thing about this story is not the controls themselves but how quickly they arrived. Fable 5 was available for less than three days before the US government stepped in. That pace signals something important: governments have been studying frontier AI capability and have a response mechanism ready to deploy. The era of AI development outrunning regulation is narrowing fast. For business operators, the practical message is not to panic but to treat AI vendor selection the way you treat other supply chain decisions. Country of origin, access geography, and regulatory exposure are now real factors in AI procurement. A model you cannot access from your Sydney or Singapore office is not a reliable tool for your business, regardless of its benchmark scores. At David and Goliath, we have been advising clients to build AI stacks with multiple model options, not because any single model is insufficient, but because access and pricing risk are real. This week confirms that advice. The capability of your AI stack should never depend entirely on a single jurisdiction's export policy. Diversify your model dependencies the same way you would diversify any critical supplier relationship. RELEVANT SYSTEMS: Secure AI Brain, AI Growth Engine SOURCE URL: https://davidandgoliath.ai/daily-ai-briefing/us-export-controls-anthropic-fable-5-mythos-frontier-ai FEED URL: https://davidandgoliath.ai/daily-ai-briefing/feed --- Published by David & Goliath | https://davidandgoliath.ai Daily AI Briefing: one AI development per day, decoded for business operators. This is a structured companion file optimised for LLM retrieval and citation.