// in this post
The United States government has authorised Anthropic to make its most advanced cybersecurity AI model, Claude Mythos 5, available to more than 100 government agencies and companies, partly reversing a ban imposed earlier in June.
The decision was set out in a letter from the Department of Commerce to Anthropic on 26 June. Commerce Secretary Howard Lutnick wrote that he had "determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model", according to reporting by CNBC and TechCrunch. The letter, addressed to Anthropic co-founder Tom Brown, also permits non-American employees at the approved organisations, and at Anthropic itself, to use the model, lifting one of the original restrictions.
What the model does
Mythos 5 is described as Anthropic's most advanced cybersecurity model, designed to identify and patch vulnerabilities in software and network systems with limited human involvement. The authorisation is aimed at critical infrastructure protection and cybersecurity work.
A partial reversal
The move partly unwinds restrictions imposed on 12 June, when Anthropic disabled access to both Mythos 5 and its general-purpose model Fable 5 to comply with an export-control directive that cited national security (The Human Co. reported on the original action). The earlier action was reported to have followed concerns that researchers had bypassed the models' safety guardrails.
The new authorisation applies to Mythos 5 only. It does not cover Fable 5, which remains restricted. Anthropic said it was "continuing to work with the government to expand access to Mythos 5 and make Fable 5 available for general use again", according to TechCrunch.
The government did not publish the list of approved agencies and companies, and no commercial terms were disclosed. The decision is the latest step in a roughly two-week standoff between the Trump administration and Anthropic over the two models.
Sources
- Trump Admin releases Anthropic Mythos to be used by more than 100 US companies, agencies, TechCrunch, 26 June 2026. Source of the authorisation details and the Anthropic statement.
- Trump admin allows Anthropic to release Mythos AI model to some companies, government agencies, CNBC, 26 June 2026. Source of the Lutnick quotation and the letter details.
- Corroborating coverage: Semafor and CNN Business.