What happened
On June 26, 2026, Commerce Secretary Howard Lutnick sent a letter to Anthropic co-founder Tom Brown announcing a 'revision to the license requirements' under the June 12 BIS export control directive. The directive partially lifts the blanket suspension of Mythos 5 and Fable 5, permitting Anthropic to restore access to Claude Mythos 5 for a vetted list of more than 100 U.S. companies (including Fortune 500 firms) and federal agencies designated as 'trusted partners.' Lutnick wrote: 'I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model.' The original June 12 export control directive — which barred 'any foreign national' from accessing either model and caused Anthropic to disable both globally — remains in force. Fable 5 (the consumer-facing variant) remains blocked with no announced timeline. The partial lift follows two weeks of high-stakes negotiations in which Anthropic worked with the US government to implement safeguards addressing jailbreak concerns raised by the NSA.
Why it matters
This is the first instance of the US government actively managing frontier AI model access as a national-security licensing regime — not a blanket ban or a blanket release, but a named-entity authorization list administered by the Commerce Secretary. It establishes a live precedent: frontier AI models with advanced cybersecurity capabilities are treated as dual-use infrastructure subject to EAR-style licensing. Organisations that are not on the 'trusted partner' list remain locked out of Mythos 5 and all of Fable 5. The framework for how partners get added, what safeguards are required, and how the list evolves is still being built in real time, creating acute compliance uncertainty for enterprises that depend on these models for security operations.
Action needed
Enterprises and government agencies currently relying on Anthropic Mythos 5 or Fable 5 should: (1) determine whether they appear on Commerce's 'trusted partner' list; (2) if not, pursue the application/vetting process through Anthropic and Commerce; (3) implement contingency AI architectures using alternative models for continuity; (4) monitor Fable 5 re-authorization timeline. Foreign nationals (including employees) remain barred from access under the still-active June 12 directive — HR and access-control reviews required.