The US directive to restrict foreign access to Anthropic's leading models signifies a change in direction.

      The US government has instructed Anthropic to prevent foreign nationals from accessing its two most advanced AI models; instead of implementing a selective nationality restriction on a shared cloud service, the company opted to disable access for all users.

      On June 12, late in the day, Anthropic deactivated Claude Fable 5 and Claude Mythos 5 worldwide, just three days after unveiling Fable 5 as its most powerful public model. This action marks what several sources describe as the first export-control measure targeting specific AI models rather than the chips or hardware that support them. The directive prohibited access for foreign nationals both within and outside the United States, making it impractical to enforce selective access on a multi-tenant service, thus leading to a complete shutdown as the easiest solution.

      According to Anthropic, the government's primary concern is a method to jailbreak Fable 5, which would allow users to circumvent the safeguards designed to prevent the model from generating harmful outputs. This decision followed a jailbreak demonstration shared on X on June 10 by a prominent individual who claimed to have bypassed the model's safety features.

      Anthropic stated that it assessed the report believed to have triggered the directive and concluded that the capability demonstrated is easily available in other models, including OpenAI's GPT-5.5. The company criticized the response as excessive, arguing that recalling a commercial model used by millions based on a narrow potential jailbreak could set a precedent that would "halt all new model deployments for all frontier model providers" if it became the standard approach.

      This situation could have broader implications across the industry as it establishes a precedent whereby the government can withdraw a launched model on national-security grounds, impacting multiple firms rather than just one.

      The jailbreak at the center of the controversy was shared publicly on X on June 10 by a well-known figure in the model-breaking community, who claimed to have defeated Fable 5’s safety measures. The government’s directive came shortly afterward, which Anthropic contests, stating that a publicly available demonstration rather than a private determination served as the basis for withdrawing a model that hundreds of millions of users rely on.

      For the administration, this action reflects a more proactive stance toward an industry it has previously engaged with, and the decision to intervene in a key product so soon after its launch is noteworthy. The nature of the directive also plays a role; since it targets foreign nationals on a shared cloud service rather than focusing on a specific export, the only feasible way to comply was to deactivate the models for all users, resulting in a global outage rather than one based on nationality.

      For enterprise customers who had developed their operations around Fable 5 and Mythos 5, the immediate challenge is significant: the models they relied on are now unavailable, and the timeline for their reinstatement is controlled by factors outside of Anthropic's influence. The forthcoming developments will depend significantly on decisions made in Washington.