The administration of US President Donald Trump and Anthropic have reached an agreement regarding the AI model Fable 5. As a result, the model’s access restriction has been removed, and since July 1 it is again available to users worldwide through the Claude Platform, Claude ai, Claude Code, and Claude Cowork platforms.

Recall, on June 12, the US authorities imposed a ban on the use of the Fable 5 and Mythos 5 models by foreign governments, companies, and individuals. To comply with regulatory requirements, Anthropic has temporarily disabled access to both models for all users, including US users.

The restrictions are based on Amazon testing. Some of the defenses in Fable 5, which are designed to block requests for sensitive topics (cybersecurity, biology), have been found to be circumvented by changing the request wording. As part of a new agreement, Anthropic has committed to address identified vulnerabilities in the model’s protection framework.

Fable 5, released in early June, was the first publicly available model of the Mythos class to have enhanced capabilities for finding vulnerabilities in cybersecurity systems. In parallel, Anthropic also restored access to the Mythos 5 model for a number of US companies after receiving approval from the US government.