Two weeks after the US government slapped export controls on Anthropic’s most powerful AI models, Mythos 5 is back — at least for a select group of approved organizations.
Commerce Secretary Howard Lutnick sent a letter to Anthropic co-founder Tom Brown on June 26th confirming a “revision to the license requirements” after the company worked with the government to address safety concerns around both Mythos 5 and the public-facing Fable 5 model.
The deal allows Anthropic to offer Mythos 5 to a curated list of cyber defenders and infrastructure providers. Non-US nationals at those approved organizations can also access the model — a key concession since the original directive blocked all foreign access, including Anthropic’s own international employees.
Fable 5, the consumer-facing model, remains in limbo with no clear timeline for public rollout.
“We are working to provision the approved set of providers and restore their access to Mythos 5 as quickly as possible,” an Anthropic spokesperson said.
The reversal came after mounting pressure. Anthropic’s competitors had been pulling ahead on cybersecurity benchmarks, and US government departments — including the NSA — had lost access to the model. Industry leaders warned that China could make AI advances while American labs were sidelined by regulation.
