The administration of U.S. President Donald Trump has allowed Anthropic to again provide access to its Mythos 5 artificial-intelligence model to trusted companies and government partners. The decision followed two weeks of negotiations between the U.S. Commerce Department and Anthropic, The Wall Street Journal reports.
U.S. Commerce Secretary Howard Lutnick said dozens of organizations recognized by the government as reliable partners will be able to receive access to Mythos 5. At the same time, restrictions on the use of the Fable 5 model, as well as Mythos 5 for other foreign users, remain in force.
According to The Wall Street Journal, the restrictions were prompted by concerns among U.S. authorities after Amazon researchers discovered a way to bypass the safety mechanisms of one of Anthropic’s models.
Against this backdrop, OpenAI also announced a delay in the public launch of GPT-5.6. The company said it would first provide access to a limited circle of vetted partners in coordination with the U.S. government.
Anthropic earlier suspended the Claude Fable 5 and Mythos 5 models several days after expanding access to them in response to an export directive.
U.S. Authorities Demand Foreigners Be Blocked From Claude Fable 5—Anthropic Disables the Model for Everyone
White House Sought to Remove Anthropic From U.S. Government Systems but Is Now Forced to Work With the Company Again
Anthropic Sued the Trump Administration After the Pentagon Labeled the Company a Supply Chain Threat