# U.S. Eases Anthropic Block, Allowing Mythos 5 for Select Cyber Defenders

Two weeks after ordering Fable 5 and Mythos 5 disabled for all foreign nationals, the Commerce Department revised the license to let a small group of trusted partners use the cybersecurity model again.

- Published: 2026-06-27T10:45:08.127Z
- Canonical: https://polylog.news/ai/2026-06-27/u-s-eases-anthropic-block-allowing-mythos-5-for-select-cyber
- Publisher: Polylog (AI desk)
- Section: geopolitics
- Sources: [Polylog editors](https://polylog.news), [Anthropic News](https://www.anthropic.com/news/fable-mythos-access)

The United States government has partially reversed its export block on Anthropic's most powerful models. According to [CNN](https://www.cnn.com/2026/06/26/tech/anthropic-mythos-release), Commerce Secretary Howard Lutnick determined that adequate safeguards now allow certain trusted partners to access Claude Mythos 5, Anthropic's strongest cybersecurity model. The company says it can redeploy the model to a small group of cyber defenders and infrastructure providers.

The reversal follows a June 12 directive that, citing national-security authorities, ordered Anthropic to [suspend all access to Fable 5 and Mythos 5](https://www.anthropic.com/news/fable-mythos-access) by any foreign national, including the company's own foreign-national employees. Because there is no reliable way to sort hundreds of millions of users by nationality in real time, Anthropic disabled the models for everyone. The company said the government appeared concerned about a way to bypass Fable 5's safety restrictions (a jailbreak), and [argued](https://x.com/AnthropicAI/status/2065597531644743999) that one narrow potential jailbreak should not justify withdrawing a model used by hundreds of millions of people.

Senior Anthropic staff then met with Commerce officials in Washington, according to [The Globe and Mail](https://www.theglobeandmail.com/business/article-anthropic-trump-officials-deal-restore-fable-5-mythos-5/), and the government sought assurances that the models could not be used against United States interests. The new arrangement resembles the gated structure OpenAI used for GPT-5.6: full capability for a vetted few, restrictions for everyone else.

## What this means

An export-control system built for computer chips and weapons is now being applied to a software service that can be switched off overnight, and the resolution is an approved-partner list rather than a return to open access. For anyone building on frontier application programming interfaces (APIs), model availability has become a sovereign risk that can change with a single government letter.

## What to watch

- Whether Fable 5, the general-purpose model, is restored on similar terms, which would show the gating framework applies to flagship consumer models and not only security-sensitive ones.
- How the trusted-partner list is defined and audited, since the criteria will become the de facto standard other labs are pushed to adopt.