Anthropic Mythos 5 redeployed to select US partners again

Anthropic's Mythos 5 has been cleared for limited redeployment after a June 26 letter from Commerce Secretary Howard Lutnick to Anthropic co-founder Tom Brown approved access for a select group of organizations, while Fable 5 remains unavailable with no timeline for a public rollout.

Who can use Mythos 5 now?

Mythos 5 can be used again only by a small set of approved organizations: trusted enterprises, US government bodies and the specific providers named by the Commerce Department, and both Anthropic employees who are US nationals and non-US nationals who are members of approved organizations. The June 26 letter says Anthropic "worked with the U.S. government to address risks" and Lutnick wrote, "These efforts have yielded significant progress," permitting "certain trusted partners to access the Claude Mythos 5 Model." Anthropic spokesperson Danielle Ghiglieri said the company had "received notice from the US government that Mythos 5, our strongest cybersecurity model, can be redeployed to a small group of cyber defenders and infrastructure providers," and that the company is provisioning the approved set of providers.

Why were Mythos 5 and Fable 5 blocked in the first place?

The models were hit by a June 12 export control order that required Anthropic to suspend access by "any foreign national" to Mythos 5 and Fable 5, a restriction that covered non-US citizens inside and outside the United States and included Anthropic employees. That order followed concerns about potential security vulnerabilities and guardrail bypasses discovered during testing. Anthropic took the models offline and spent two weeks in negotiations with the administration to resolve those concerns. The Verge has previously reported that a vulnerability related to how the models handled certain prompts — a jailbreak that affected defensive security workflows — helped trigger the fast-tracked government action.

What changed in negotiations and how does this compare to other labs?

The Commerce Department kept the original export control directive in place but granted an exception for Mythos 5 similar to the limited-preview approach used for OpenAI's GPT-5.6, which was announced earlier the same day as Anthropic's approval. Under the exception, members of approved organizations who are not US nationals are cleared to access Mythos 5. Anthropic says it will continue working with the government to expand access and to make Fable 5 broadly available again, but it has not given a timeline. The negotiations replaced Anthropic CEO Dario Amodei with co-founder Tom Brown as the lead contact with the administration, while the company works on protocols, standards and releases for Mythos-class models as requested by the government.

Why it matters

Restoring Mythos 5 to a confined set of cyber defenders reduces immediate capability gaps for US infrastructure protection and for government cybersecurity teams that had lost access. The restricted-availability route lets the government retain control over who can use the most powerful defensive model while keeping Anthropic partly in the commercial market. The stalled status of Fable 5, however, maintains uncertainty for enterprises and could affect Anthropic's revenue plans; previous reporting notes Mythos-class models were expected to be a high-margin product for Anthropic and the company has large compute commitments, including a reported SpaceX deal worth $15 billion per year for data center access.

What to watch

Watch for any official expansion of the approved-provider list and for a timeline from Anthropic on Fable 5 availability, and monitor whether the Commerce Department amends "all other requirements of the June 12 letter," which Lutnick said remain in effect until further notice. A wider reopening, or a repeatable licensing process for future releases, would be the clearest sign the administration is moving from case-by-case exceptions to a longer-term framework.