US lifts partial ban on Anthropic's Claude Mythos AI

In early June the U.S. Commerce Department issued an export ban on Anthropic’s AI models Claude Mythos 5 and its commercial variant Fable 5, citing a technique that could bypass safety filters and potentially enable large‑scale vulnerability exploitation. The ban forced Anthropic to shut down both models worldwide, restricting access for all foreign users, including its own staff.

After roughly two weeks of negotiations, Commerce Secretary Howard Lutnick sent a letter to Anthropic on 27 June stating that “appropriate safeguards have been put in place.” The letter permits a limited group of trusted partners to regain access to Claude Mythos 5, but the broader Fable 5 model remains blocked. Anthropic challenged the decision, arguing the bypass applies only to a narrow scenario and noting that other leading models, such as OpenAI’s GPT‑5, have similar capabilities without comparable restrictions. More than 100 cybersecurity experts and executives from firms like Adobe and Nvidia signed a joint letter opposing the ban, warning it could hinder defensive security work. The controversy also spurred some users toward Chinese AI providers such as DeepSeek, which raised a record $7.4 billion funding round during the ban period.