U.S. Lifts Export Controls on Anthropic AI Models

On Tuesday, AI company Anthropic announced that the U.S. government had lifted export controls on its newest models, following a more than two-week suspension prompted by cybersecurity concerns.

Anthropic said that Claude Fable 5 would be available to consumers worldwide starting on Wednesday, although use of the more capable Claude Mythos 5 will remain limited to certain approved organizations. Access to both models had been blocked since June 12, when the Trump Administration imposed restrictions on their use by foreign nationals.

The White House was concerned that Anthropic’s safeguards could be bypassed to help identify software vulnerabilities, allowing for disruptive cyberattacks at potentially large scale. In response, Anthropic said it had strengthened its models’ safety systems, and committed to continuing cooperation with the U.S. government on testing, standards, and threat reporting.

Anthropic competitor OpenAI also rolled out its latest model, GPT 5.6, with a limited release to comply with U.S. government requests.

Prior to this, the Administration had taken a “light touch” approach to AI, seeking to federally preempt state regulation of the technology and issuing an executive order that eschewed a mandatory licensing regime in favor of a voluntary approach. However, the increasing sophistication of new AI models may be prompting a change of course.