Commerce Secretary Howard Lutnick lifted the Bureau of Industry and Security’s export controls on Anthropic’s Claude Mythos 5 and Fable 5 on June 30, ending an 18-day shutdown that had become the first live test of Trump’s new frontier-AI framework. Licensing requirements on exports, reexports, and transfers are gone. Anthropic said Fable 5 will return to AWS, Google Cloud, and Microsoft Foundry “as soon as possible.”

The standoff traced a fast arc. On June 2, Trump signed the “Promoting Advanced Artificial Intelligence Innovation and Security” order, establishing a voluntary 30-day pre-release government access window for covered frontier models. Earlier drafts had set that window at 90 days. On June 12, BIS suspended all foreign-national access to Mythos 5 and Fable 5 on national security grounds, citing what Anthropic characterized as a narrow jailbreak finding. The company disabled the systems and pushed back publicly: “We disagree that the finding of a narrow potential jailbreak should be cause of recalling a commercial model deployed to hundreds of millions of people.” A senior White House official told Politico the controls were “a last resort.”

By June 26, Lutnick had authorized Mythos 5 for select U.S. organizations and federal agencies under “appropriate safeguards,” in a letter addressed to co-founder Tom Brown, who had replaced CEO Dario Amodei at the negotiating table. Four days later the controls collapsed entirely. Anthropic’s Glasswing cybersecurity program is being expanded, with federal recipients selected under the Section 3 “trusted partners” mechanism.

The template is now legible: voluntary access, trusted-partner gating, Commerce as the choke point. A classified NSA-led benchmarking process, due within 60 days of the order, will define which future models trip the same wire.

Sources

Sources