US Lifts Export Controls on Anthropic AI Models

The U.S. Department of Commerce lifted export controls on June 30, 2026, allowing Anthropic to restore global access to its Claude Fable 5 and Mythos 5 AI models starting July 1. This ended an 18-day suspension triggered on June 12 after reports from Amazon researchers revealed that Fable 5's safeguards could be bypassed to identify software vulnerabilities and produce exploit code. To comply with the initial ban on foreign nationals, Anthropic had disabled both models globally.

To secure the rollback, Anthropic implemented a new safety classifier that blocks the identified jailbreak technique in approximately 99% of cases. The company also agreed to proactively detect security risks, report malicious activity to the government, and provide federal agencies with pre-release model access for evaluation. While Fable 5 returned to general public availability, Mythos 5 remains restricted to a select group of over 100 trusted U.S. organizations and critical infrastructure providers via Project Glasswing.

These actions follow a June executive order by President Donald Trump establishing a voluntary framework for the government to vet frontier AI models for national security risks for up to 30 days before public release. This oversight extended to OpenAI, which limited the rollout of its GPT-5.6 Sol model to a small group of government-approved partners.

In response to the regulatory environment, Anthropic launched Claude Sonnet 5, a mid-sized model designed to avoid the high level of scrutiny applied to frontier models. The company is now partnering with Amazon, Microsoft, and Google to create an industry-wide consensus framework for assessing AI jailbreak severity.