Claude Mythos 5 available again, but only for select organizations

The U.S. government has partially lifted restrictions on AI company Anthropic, granting access to its most powerful model, Claude Mythos 5, to more than 100 select organizations, primarily companies and institutions managing critical infrastructure. This move comes after an initial ban two weeks prior, imposed due to national security fears that advanced AI models could fall into the hands of adversaries like China or Russia and be weaponized for cyberattacks.

The government's previous decision prohibited access to Mythos 5 and Fable 5 for all users. Similar restrictions are in place for OpenAI, which has delayed the public release of GPT-5 at the U.S. government's request, making it available only to verified partners for now. The new decision means only pre-approved organizations can use the Mythos 5 model.

The U.S. Department of Commerce has waived the export license requirement for employees of these organizations, even if they are not U.S. citizens. However, this selective access has sparked criticism. It remains unclear what criteria the government uses to choose which companies receive access to the most advanced AI models. OpenAI CEO Sam Altman has voiced support for security vetting of AI models but opposes the government dictating who can use them.

Anthropic is continuing discussions with U.S. authorities regarding broader accessibility for Mythos 5 and the public release of Fable 5. Experts warn that prolonged restrictions on releasing new AI models could risk China narrowing the technological gap with the United States in the field of artificial intelligence.

“

I support security vetting of models, but not the government deciding who gets to use them.