← All resources
Tech & TrendsJuly 2026 · 4 min read

U.S. allows Anthropic to release Mythos AI to ‘trusted’ US organizations

What actually happened with Anthropic's Mythos AI? On 26 June 2026, the US Commerce Department lifted a two week export block and allowed Anthropic's Claude Mythos 5 model to reach more than 100 US or

What actually happened with Anthropic's Mythos AI?

On 26 June 2026, the US Commerce Department lifted a two week export block and allowed Anthropic's Claude Mythos 5 model to reach more than 100 US organisations, including Fortune 500 companies and government agencies. Commerce Secretary Howard Lutnick said "appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model."

The original suspension followed a reported jailbreak vulnerability and concern that Mythos 5 had reached partners with possible China links, including a South Korean telecoms provider. The same day, OpenAI released GPT 5.6 to approved US government partners under a similar review process, which points to a new pattern: US regulators now actively gatekeeping which organisations get access to frontier AI models, not just how the models behave.

Is Mythos 5 something my business can use?

No. Mythos 5 is a specialised, dual use model built for cybersecurity vulnerability research and biology research, not general business tasks, and access still requires vetting under Anthropic's trusted access programme. Pricing reflects that: $10 per million input tokens and $50 per million output tokens, far above Anthropic's standard consumer and business tier pricing.

For everyday use, Anthropic's answer is Claude Fable 5, which runs on the same underlying model but with safety routing that redirects risky queries. Fable 5 remained blocked from wider release as of the announcement, so the tool most small businesses would actually touch is not yet fully out.

Why should a UK small business owner care about a US export ruling?

You won't ever license Mythos 5, but the story matters because it shows AI access is now a regulated, permissioned thing, not a simple sign up and use product. That has two direct consequences for anyone relying on AI to be found, recommended, or to run their operations.

First, the tools you depend on can be paused, restricted, or re priced with little warning, exactly as this two week block demonstrated. Second, and more relevant day to day: being recommended by AI assistants like ChatGPT, Gemini and Google AI Overviews is becoming the new local search engine optimisation, and that recommendation layer is being built on infrastructure you do not control and cannot see inside.

What's the practical takeaway for online founders this week?

Don't chase Mythos 5 headlines, focus on making sure your business is actually visible to the AI tools you can already use. Concretely: claim and verify your Google Business Profile, keep opening hours and services current, and publish a real website with clear, factual service pages rather than relying only on a Facebook page or a Linktree.

  1. Verify your Google Business Profile this week if you haven't in the last three months. One Leeds salon went from no online presence to over 40 monthly calls from Google Maps within three weeks of doing exactly this.
  2. Replace static documents (PDF menus, price lists) with live web pages. A Manchester restaurant saw online orders rise 34% within four weeks of swapping a PDF menu for a live one, partly because live text is what AI assistants can actually read and cite.
  3. Audit which parts of your business run on rented platforms you don't own, and price what owning that layer would cost instead.

Braynex Services' point of view

News like this confirms something we've said for a while: rented platforms, whether that's Facebook, Fresha, Linktree or Yell, are a liability, not a shortcut. You don't own the customer relationship, the data, or the terms, and this Mythos 5 episode shows that even the biggest AI providers can have access switched on and off within weeks by a regulator.

The businesses that come out ahead won't be the ones using the newest model. They'll be the ones who own their own infrastructure, a real website, their own booking system, their own customer data, so that whichever AI assistant a customer asks, there's something solid for it to find and recommend. We've seen this directly: a nail salon that moved off Fresha's roughly £1,800 a month commission model to its own £35 a month booking system now saves around £21,000 a year, and keeps every customer record it generates.

If you want a clear picture of where your business is exposed, rented platforms, invisible missed calls, old customer data nobody follows up, book a free audit at braynexservices.com.

Want this built for your business?

We build the digital infrastructure behind everything you read here. Book a short scoping call and we'll show you what to fix first.

Book a scoping call →