The U.S. government has dropped a requirement that forced Anthropic to obtain a special license before shipping its Mythos and Fable AI models overseas. The restriction had effectively cut off public access to what many consider the most advanced AI systems ever released. Anthropic said it will restore access starting July 1.

The export rule, imposed on June 12, required special approval before any foreign national could use the models. Anthropic found compliance impractical at scale, so it shut down public access entirely. After weeks of talks, Commerce Secretary Howard Lutnick announced that Anthropic has agreed to proactively detect security risks, work with the government on release protocols, and report any malicious activity. Critics note that Anthropic had already volunteered to do most of this months before the rule existed, leading cybersecurity experts to view the ban as less about security and more about leverage—a way for the Trump administration to punish the company for its executives’ public criticism of how the government might use AI.

Mythos was initially released in April to a select group of organizations to address concerns about its ability to exploit software vulnerabilities. A version called Fable followed in June with added guardrails. But as Asian competitors like Fugu and Tulonfeng released models approaching Mythos-level capabilities, pressure mounted on Washington to ease restrictions to keep American AI competitive. Last week, Lutnick cleared Mythos for release to White House-approved customers—similar to how OpenAI’s latest models were distributed to Trump-approved groups rather than the public. The administration’s unpredictable AI policy has left the industry uncertain about future rules, and an executive order in June calling for pre-release model review drew sharp criticism from analysts like Dean W. Ball, who recently joined OpenAI’s policy team.

Source: TechCrunch