Anthropic Restores Mythos 5 Access After U.S. Government Approval

Anthropic has announced that it is once again rolling out access to its Claude Mythos 5 AI model after the U.S. government approved a limited restoration. The move follows the model’s abrupt shutdown earlier this month and marks the first step toward making the company’s most advanced AI available again, although only to a carefully selected group of organizations.

According to Reuters, the restored rollout remains highly restricted and focuses on U.S. organizations responsible for operating and defending critical infrastructure. Anthropic emphasized that the decision does not signal a full public return for its latest frontier AI models.

Mythos 5 returns for approved organizations

The company said access is currently limited to selected U.S. organizations that support critical infrastructure. Anthropic added that it continues to work with the U.S. government to gradually expand availability over time.

The public-release version of the model, known as Fable 5, remains unavailable. Anthropic said it is actively working to restore access to Fable 5 as well, but has not provided a timeline for when the model will become available again.

The company stressed that the latest approval only applies to the restricted Mythos 5 rollout and should not be interpreted as a broader reopening of access.

Commerce Department reportedly eased restrictions

Reuters reports that the decision followed a letter issued by the U.S. Commerce Department. According to the report, more than 100 companies and institutions could become eligible to use Mythos 5 under the updated policy.

The reported guidance also removes the requirement for export licenses covering approved companies’ non-U.S. citizen employees. Reuters added that the exemption extends to Anthropic’s own non-U.S. citizen employees as well.

Organizations outside the approved group remain subject to the existing restrictions, meaning Mythos 5 continues to operate under a tightly controlled distribution model.

AI companies continue controlled rollouts

Anthropic is not the only AI developer limiting access to its newest models.

OpenAI recently announced that its GPT 5.6 family would also launch with restricted availability. Initial access is limited to trusted partners and selected organizations, with broader availability expected later this year.

For now, Mythos 5 remains available only to approved U.S. organizations involved in critical infrastructure and related sectors.

In separate news, Anthropic recently accused Alibaba of attempting to distill its AI models, adding another chapter to the increasingly competitive race between major AI developers.