Trump administration lifts ban on Anthropic’s Claude models

The Trump administration lifted restrictions on Anthropic’s latest AI models, allowing Claude Fable 5 to return to broad public availability while restoring Mythos 5 access only to a set of U.S.-based organizations approved by the federal government.

The Commerce Department blocked foreign nationals from using both models on June 12, and Anthropic withdrew the models days after their launch. Anthropic wrote that the government’s review followed a report from cybersecurity researchers at Amazon Web Services, its main cloud provider, who “had found a method of bypassing Fable 5’s safeguards” that could enable users to discover and potentially exploit software vulnerabilities.

Anthropic had warned earlier that Mythos is particularly capable of identifying software flaws that could be weaponized by malicious actors and threaten critical computer networks. Access to Mythos 5 is limited to a group of domestic organizations that received federal clearance.

OpenAI also limited release of a new model, GPT-5.6 Sol, making it available only to government-approved customers for a temporary period at the request of the administration. President Donald Trump signed an executive order last month that creates a process for federal review of advanced AI systems for up to 30 days before public release; the review framework is still being developed and developer participation is voluntary.

Anthropic did not provide a timetable for broader reinstatement of Mythos 5 access. The Commerce Department did not immediately comment on lifting the restriction for foreign users.

Officials and developers are assessing security implications as AI systems can automate tasks such as code analysis and vulnerability discovery.