U.S. partially lifts Anthropic ban on Claude Mythos 5 access

Washington partially lifted its ban on Anthropic’s Claude Mythos 5 on Friday, allowing the model back to more than 100 companies and institutions, including many Fortune 500 firms, after a two-week freeze tied to national-security concerns, but only if they were on the government-approved list.

The U.S. government issued an export control directive on June 12 to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign-national Anthropic employees. That order cut off the company’s most advanced systems just days after Mythos 5 was set to roll out through Project Glasswing as an upgrade to Claude Mythos Preview, initially to a small group of cyberdefenders and infrastructure providers.

In the letter, Commerce Secretary Howard Lutnick wrote that Anthropic had worked with the government to address risks tied to the covered models and that the effort had produced significant progress. He also wrote that an export license would no longer be needed for Mythos 5 when it was used by approved companies and certain foreign-national employees. The status of Anthropic’s other model, Fable 5, remained unclear.

Officials worried that advanced models could be useful to military-intelligence users in China, Russia or other countries seen as sensitive by Washington, and the new access rules tied access to trusted partners rather than broad commercial release.

Project Glasswing is expanding to trusted security teams and partner organizations in power, water, healthcare, communications and hardware, with some of those codebases capable of affecting more than 100 million people if attacked. During testing with U.S. intelligence agencies, Mythos 5 identified vulnerabilities in highly sensitive government computer systems.

The company confidentially filed a draft Form S-1 with the Securities and Exchange Commission on June 1, raised $65 billion on May 28 at a $965 billion post-money valuation, and announced a Seoul office and new partnerships across the Korean AI ecosystem on June 17.